A Personal Story Illustrating the Need for Improved Healthcare through Technology
I remember from my own experiences with my father after his stroke what it was like shuttling him from specialist to specialist in an effort to get him the care he needed, and I wish that better remote healthcare technologies were available during that time. Although medicine and healthcare information technology is the field in which I earn my living, there’s always something educational about firsthand experience. I am quite familiar with the field of healthcare information technology, its foibles, its benefits and its potential impact on healthcare delivery. I ran a critical care product line for a large healthcare information technology vendor and bioinformatics research for another. However, the experience one gains in actually participating and operating in and around the healthcare system in the United States is one that many of us have, many of us curse, and many of us appreciate. I’d like to direct attention to several aspects of the healthcare system that we, as Americans, may consider to be mundane. However, perhaps after reading this those so inclined might have their interests piqued and pay attention somewhat more acutely to these rather mundane items and this in and of itself may cause you to re-think how improvements in the system could benefit you and your families.
Experiences in Healthcare, and the Impact of Information Technology
We are all familiar with visiting a physician office for the first time. You know the routine: you are handed a clipboard containing half-a-dozen pages or so in which you must disclose every torrid aspect of your life. You must hand over your insurance card so a copy can be made and you must list every medication, every tablet, every vitamin you take in quantity, type, label so that THIS physician has a clear picture of who you are and what has ailed you. Some people are more organized: they make copies of this information and merely re-copy onto the forms so that they do not have to make up their stories out of whole cloth. Others are not so organized. Regardless, this information is normally maintained by your primary care physician (PCP) and is seldom shared unless you explicitly ask for records or, perhaps, your PCP is one of the more “progressive” types using an in-house electronic medical record system. Alas, currently only a relatively small number of U.S. physicians make use of health information technology. The estimate is that 17 percent of U.S. physicians and between 8 and 10 percent of U.S. hospitals employ health information systems in the form of electronic medical records for capturing and maintaining patient medical data (David Blumenthal, 2009).
However, consider having to repeat this process for each physician you visit. This takes me back to the story of my father. My father, who passed away last August, had a stroke in 2003 at the age of 85. I’ll save the experience of the basic medical challenges for another article, but I will relate that the treatment process involved half a dozen specialists and allied health professionals. All of these individuals required similar information regarding his health and history. Each one of them required the information in a format similar to the method described above with clipboard and pen.
Medical Device Integration and Healthcare Information Technology
Positive Health Wellness has a great article entitled “8 Ways Technology is Improving Your Health.” Key among the identified benefits include faster and more accurate diagnoses, better treatment options, access to and recording of data in real-time, as well as recording of data on chronic conditions in the home. The ability to record data both in the hospital and outside in an ambulatory setting are expanding and becoming more commonplace. But, the data transfer mechanisms are still improving. Let’s consider for a moment an alternate method for the requisite “data transfer” experience described in the paragraph above. Suppose that my father’s history, medications, allergies, treatments, etc. were all contained on a single “device,” such as something having the form factor and function of a Universal Serial Bus (USB), or memory, stick. Then, if each physician and specialty practice had the capability of reading such information from this device into an electronic record that employed common interfacing and formatting so that the information could be populated in a way that would be visible and accessible to each physician, a number of benefits would have resulted. First of all, the mere physical act of copying the same information over and over would not have been required. Secondly, the likelihood of inaccurately entering information could have been avoided. Given the fact that I was the primary source for data entry, I can attest to the fact that I am error-prone! Thirdly, having a complete and accurate list of his medications, his treatment plans, clinical notes, and orders all available would have provided to each specialist a comprehensive understanding of his history. This would have enabled each of them to communicate more effectively to determine how best to treat him without having to ask redundant questions of both him and me during the visit—more time could have actually been spent in treatment!
The capability and benefits described above are not out of our reach. The technologies exist to enable the scenarios described above. The benefit to patients is obvious, as can be discerned from even this simple telling. Improving healthcare delivery can be achieved without bringing rocket science to the practice—we can begin simply by doing what we currently do more efficiently and by bringing some good sense to the practice. We as citizens think nothing about going to a store and using a credit card to pay for goods and services. Yet, we have nothing equivalent in standard practice that allows us to treat the most precious good: our bodies. In the coming weeks I hope to expand upon this theme and raise awareness on the benefits of information technology and its benefits to healthcare.
One of the chronic ailments that my father suffered from was the wet (or neovascular) form of macular degeneration. Neovascular macular degeneration affects about 10% of those patients who suffer from macular degeneration in general . While there currently is no means of stopping or reversing the effects of macular degeneration, certain therapies (laser photocoagulation) can stem the bleeding associated with the wet form of the disease.
The effects on my father were heartbreaking in many ways. In my father’s working life he was a writer and editor for a number of industries, including the New York Medical Society, Ford Truck Times Magazine, and he was an advertising executive back in the 60s at J. Walter Thompson advertising as well as having his own advertising agency in the 1950s. In summary, my father’s sight was key to his livelihood. This was a man who used to read the New York Times cover to cover almost daily. In the last 5 years of his life, as a result of this ailment and the stroke he eventually suffered, he was left sightless and unable to enjoy the one thing that truly gave him pleasure.
I’m certain that many of us have equally poignant stories. During the 5 year period of both chronic and continuing medical care my father required I spent a great deal of time running him from specialist to surgeon to primary care physician to therapy and back again. I recall very vividly having to run him into Philadelphia from his home in the suburbs—about a 60-70 minute drive—for the purpose of having his eye surgeon review his progress. I remember how stressful the situation used to be: it was a fairly major production getting him out of the house and driving him down and back as he required assistance due to limited mobility. Oftentimes the visits were merely checkups of no more than 5-10 minutes duration. It was at these times that I used to ponder whether having a remote video and picture taking capability could have accomplished precisely the same thing: if his surgeon had the ability to review a photo remotely, my father could sit in the comfort of his home and have a retinal camera that I or another care giver could use to take a picture of his retina which could then be transmitted and reviewed by the surgeon remotely. Then, the visit could have been accomplished through a telecommunication session, in which the surgeon could speak with him directly over the telephone while reviewing the image. This would provide context for the imagery as well as provide for a much less stressful environment for my father.
Recently, Healthcare Information Technology (IT) News reported “remote monitoring not only saves unnecessary trips to the emergency department, but prevents readmissions to the hospital” . Unfortunately, the same article reports “healthcare payers are resistant to providing reimbursement for remote patient monitoring.” A chief reason for this seems to be the fact that the payer-provider reimbursement model is not adequately structured to take advantage of the benefit.
It would seem to me that the use of the healthcare information technology would reimburse itself. Ignoring the time spent in traveling to and from the surgeon’s office, consider the fact that the visit itself could be shortened and accommodated on a schedule that could make most effective and efficient use of both parties: patient and provider. For example, a virtual office visit could be held at any time during the day (not just during “normal” office hours) and could even be managed from the provider’s home office. Of course, key to this would be the availability of a patient record in which information could be securely uploaded (e.g.: retinal imagery). A personal health record could have served this purpose. Furthermore, the relaxed setting of the patient’s home would have enabled a much more relaxed environment for the patient.
While the scenario I have described is not unique, it serves to illustrate a broader need and provides a compelling motive for healthcare information technology delivered through telehealth and telecommunication. By linking healthcare information technology with existing means for communicating over telephone lines it is possible to achieve ends that will ultimately benefit chronically and elderly patients. In the next installment, I will address the benefits for other diseases, including stroke and glucose, and how the case for healthcare information technology has real benefits for the homebound or chronically ill patient.
John R. Zaleski Comments on the Common Device Connectivity AHIC Document
Some comments are offered on the functional requirements associated with high-acuity medical device communication in relation to the American Health Information Community (AHIC) priorities on common device connectivity (CDC) to electronic health records (EHRs), as pertains to the U.S. Department of Health and Human Services ONC AHIC document, title shown in figure above.
The US Department of Health and Human Services Office of the National Coordinator for Health Information Technology has chartered the development of a series of documents to represent American Health Information Community (AHIC) priorities for national health information activities. The 2009 Common Device Connectivity (CDC) Extension/Gap document, as requested by AHIC, was commissioned to address information transfer from “high-acuity and inpatient diagnostic/therapeutic medical devices…into electronic health records.”
ONC Common Device Connectivity, Defined
As stated in the scope of the ONC CDC AHIC Extension/Gap, section 2.2, Common device connectivity is
“…the means by which high-acuity and inpatient clinical device information such as settings, measurements, and monitoring values are communicate to and from [the electronic health record] and other specialized clinical information systems.”
Examples cited include vital signs monitors, mechanical ventilators, anesthesia, and infusion pumps. Radiological devices are explicitly excluded from consideration. As such, single- and multi-parameter data from such devices are assumed the primary sources of data for communication to EHRs and clinical information systems (CISs).
It is noted, as stated within section 1.2 of the CDC AHIC Extension/Gap document, that as of the publication date of this CDC document, available at
“…has not formally addressed the interoperability considerations for connectivity between medical devices and EHRs.”
Progress has been made through the spring and summer of this year via the Tiger Team efforts initiated earlier this year, with focus on the Remote Monitoring Use Case. The 2008 Remote Monitoring (RMON) Use Case highlights communication from ambulatory settings to EHR and PHR. The functional requirements associated with CDC related to communication and exchange of information between medical devices and the electronic health record are highlighted in section 3.0 of the subject document. As is pointed out in the preamble to section 3.0, it is implicit in these functional needs that what are described are key capabilities and not detailed, explicit functional requirements, representative of those expected in a mission, software, and interface requirements specification. Rather, the focus is on the high-level functional needs.
Approach to Device Connectivity Functional Needs
My approach is to list each of the explicit functional needs (as quoted directly from the document—in italics) and then provide my feedback on the potential implications of each.
Zaleski Comments on ONC Common Device Connectivity
3.0 Functional Needs
A. The ability to configure and register a device to communicate with an EHR or other system.
i. When a device is set up within an organization to communicate measurement information, the device is configured and registered within the organization’s electronic health record to uniquely identify the device and enable connectivity between the device and system.
Zaleski Comment 1: Medical device is registered within EHR
This is usually a manual process today. The ability to associate a device with a patient is normally managed through the clinical software. One method is through the manual assignment via the clinical information system, such as through the critical care flow sheet. A pick list may be shown in which a clinician assigns the available devices to patients. Other approaches to associating medical devices with patients are being evaluated by various interoperability vendors, including the automated association via barcode or radio frequency identification token through a common device token, similar to a serial number.
B. The ability to associate patient identification and device information with an EHR.
i. Patient registration, location, and identification information available within the EHR is uniquely associated with the patient’s monitoring device using standardized mechanisms for admission, transfer, and discharge from beds, units, wards, and entities within the facility.
Zaleski Comment 2: Uniquely associating medical device data with a patient
Many vital signs monitoring systems provide methods for associating monitors in individual patient rooms with patients within a flow sheet, in the format of a bed board. Patients are assigned by nursing upon admission to the unit. The key identifying information can include a medical record number and visit identifier. Upon discharge, patients are disassociated using the bed board mechanism again. This process, while manual, addresses the point identified above.
ii. In the event patient identification information is associated with a device in error, the device can be disassociated with the current patient within the EHR and associated with the correct patient.
Zaleski Comment 3: Disassociating a medical device and device connectivity data with a patient
Some clinical information systems (CISs) today provide the capability to disassociate the medical device from the patient through a flow sheet user interface. Those CISs that provide the equivalent of “bed boards” whereby patients are manually assigned to monitors (in rooms) via this user interface is one mechanism by which this can be accomplished.
iii. A patient may be placed on a monitoring device prior to the completion of patient registration or the availability of patient identification information within the EHR, especially in emergent or critical situations. The measurement information is available in the EHR upon initiation of the monitoring function or medical device initiation, and can be reconciled with patient registration or patient identification information within the EHR when available. Data collected prior to patient registration should be buffered and retained for a reasonable period of time sufficient to complete the registration process.
Zaleski Comment 4: Merging data collected through device connectivity with a patient after measurements begin
Again, certain CIS flow sheets support this, with association or linkage to patient done after admission to the unit. The HL7 admission, discharge and transfer (ADT) messages arriving from an existing master registration system to the unit can then be used to link the patient-specific identifying information to the vitals data measured from the bedside equipment. Once linked, the observations and measurements can be sent back to electronic health records.
iv. Organizational policies and procedures may require medical device measurement values within a patient’s record to be validated by a licensed clinician prior to being stored within a patient’s record. This function may prevent the charting of erroneous values within a patient’s permanent medical record.
Zaleski Comment 5: Medical devices should communicate all data as part of device connectivity function
The validation step is key to ensure that the data are indeed a true and accurate representation of the measurements from the patient. Furthermore, context added to the measurements (for example, clinical notes or text) that establish conditions at the bedside that may impact or influence the measured observations are also critical and necessary to communicate to the electronic health record.
C. The ability to associate patient identification and device information with an EHR.
i. Measurement and device information generated by the medical device is communicated to the EHR. Measurement information such as device settings, parameters, values, and units may be utilized by the EHR and/or clinical decision support (CDS) systems to support patient management.
ii. The devices should communicate state, error conditions, and user selections to support the analysis of adverse events.
Zaleski Comment 6: All measurements or findings obtained through device connectivity should be associated with a patient and collected through the EHR
This causes me to think about the work I conducted when I was at PENN in the early 90s, and why I became involved in the field of medical devices and clinical decision making in healthcare. Having a complete and accurate record of the settings, parameter values, error conditions, etc. is certainly important for documentation purposes. But, moreover, it is essential to the “art” of making clinical decisions in an advisory role to the bedside clinician. What also speaks to me here is the necessity to begin thinking in terms of real-time management and monitoring of data—through the electronic health record! Again, it is certainly necessary to have a complete and accurate record of information from the documentation perspective. However, we should bear in mind that manual recording of this information has been done for decades. Presumably, benefits in terms of reduced errors, improved quality, and interventional guidance can be offered to the clinician by monitoring and recording this information in a timely manner. It would seem to follow that as much of this information is available in real- or near real-time, it would be beneficial to the patient to record information in as high a frequency as possible and practicable. Furthermore, status and error information that are logged could be made available to biomedical and IT departments for servicing and quality control purposes.
D. The ability to support point-of-care integration to uniquely identify a device and related components, communicate device setting and detailed device information, associated with each measurement value, to the EHR.
i. When a patient device is replaced by another device of the same type, measurement information may seamlessly populate the EHR. The devices may be from different manufacturers, but communicate the same information to the EHR. The EHR recognizes the measurement parameters and is able to represent the measurement values consistently within the EHR. Device information, settings, and metadata specific to each device is associated with each measurement value and is accessible within the EHR. This is accomplished via a standards-based first communication link interface between the point-of-care device and the EHR, device intermediary, or device gateway.
ii. A patient placed on multiple monitoring and patient care devices that need to be associated with the patient within the EHR. When multiple devices are capturing the same measurement or monitoring parameter, the information available within the EHR enables clinicians to distinguish between the measurements and determine the measurements that are captured from each device.
iii. Device data should be uniquely associated with the device, the patient, and the date and time the data was acquired, sent, and received.
Zaleski Comment 7: Devices of similar function should have data of similar or precisely the same character
Standards-based communication from instrument or device gateways is typically accomplished using an HL7 result transaction. While the specific segment syntax can vary depending on the peculiarities of the device and the manufacturers’ objectives, this is more often than not the case. When device gateways do not exist, then the form of communication can be rather proprietary. Those in the device community are engaged in a continuing dialog on how to address this situation. Yet, from a clinical perspective, if two devices are interchanged (measuring the same parameter), then it may be in the interest to note the change in device as variations in device sensitivity, behavior, and manufacturing may result in some slight variation or difference in reported output. Yet, such variation should be well within the range of clinical significance so as not to raise a question as to the veracity of the result. Furthermore, one poor practice I have seen is representing two of the same values as two separate entries in flow sheets. For example, oxygen saturation from two different SpO2 cuffs. If these represent the same value (and not, for example, SpO2 and SaO2), then the values shown multiple times or presented in parallel with one another can cause confusion. While certain measurements can vary depending upon where measured (left arm versus right arm blood pressure measurement), and the differential is indeed necessary for clinical decision making, diagnosis and treatment, care must be taken so as not to present redundant measurements before the eyes of the clinician that may in fact be the same in every respect except for name. This simply will serve to confuse.
E. The ability to communicate measurement intervals and device setting information within the EHR.
i. When a patient is placed on a medical device, the clinician’s order details may specify measurement intervals for patient information to be communicated to the EHR.
ii. Depending upon patient acuity and monitoring needs, measurement intervals may need to be modified during the course of care. A clinician may modify the measurement parameters and intervals via the EHR or by modifying the device directly. Measurement interval information is communicated from the device to the EHR so the clinician may access this information.
iii. Inbound device settings and controls from the EHR may be subject to clinical oversight, validation and verification at the point of care prior to execution on the instrument itself.
iv. Measurement intervals are reconciled against the system time available from the EHR to ensure consistent and accurate identification of time intervals in absolute time.
v. The communication of multiple interval types should be supported (e.g. episodic, regular, quasi-continuous, sampled waveform, continuous waveform).
Zaleski Comment 8: Measurement intervals as requirements for device connectivity integration and ordering by clinicians
An example that is used in clinical practice is the ordering of initial support on mechanical ventilation upon admission to an intensive care unit (ICU) of a post-operative coronary bypass graft patient. For instance, upon admission, initial ventilator settings of intermittent mandatory ventilation (IMV) at 12 breaths per minute, with a forced inspired oxygen of 100% and a positive end expiratory pressure (PEEP) of 5 cmH2O might be ordered. Then, as the patient is weaned down, the order is changed over time.
F. The ability to query the device or device intermediary [Zaleski adds: I interpret this as the device gateway] for additional information captured by the device that may not have been communicated to the EHR.
i. A clinician may request certain intervals for viewing device measurements or information within the EHR. If a patient event occurs that requires further investigation, the clinician may utilize the EHR to query for additional retrospective device information or measurement details that were not initially communicated to the EHR based upon the data intervals set for the patient.
Zaleski Comment 9: Ability to drill into data collected through device connectivity for retrospective assessment
Clinical information systems that support automated collection at the bedside typically display automated vital signs information in discrete intervals. These intervals can vary. Typical ranges are one set of parameters every several minutes to once an hour, with typical values being in the quarter-hour range (i.e., once every 15 minutes, or q15). Flow sheets and their supporting medical device interoperability software need to allow for more or less frequent collection of information. The challenge remains that unless medical devices at the point of care provide for local storage of their data (possibly through their intermediaries), there may be no possibility to recall retrospective data on a patient.
G. The ability to communicate device and measurement information to the EHR when there is a lapse in EHR connectivity.
i. If a break in network connectivity occurs, or other factors prevent device communication to the EHR, device and measurement information is communicated to the EHR when connectivity is restored. Upon establishing or re-establishing this connectivity, there is no loss of measurement information in the EHR. In addition, details associated with measurement or device settings are communicated with the appropriate timestamp and patient parameters (e.g., identification, device settings) present at the time of information capture at the device.
ii. A notification may be sent to the EHR notifying of the event in which data transmission or communications are lost between the EHR and medical device. This notification consists of a standard health and status message that confirms device connectivity and general operation.
Zaleski Comment 10: Caching of data through device connectivity when gaps in networking or physical connectivity occur to EHR
A necessary requirement and cannot be overemphasized. Quality of Service undergirds this. Assured delivery of medical device data must occur if we are to use such data for intervention. As manufacturers of medical devices evolve more towards a plug and play paradigm, perhaps analogous to USB 2.0, this will assist in achieving this requirement. What we are talking about here is intelligent connectivity: devices “know” to whom they were attached; their data are not lost in the event of inadvertent loss of connectivity; the data can be picked up from where it was lost upon reconnection. Some monitoring systems provide what is typically called a “full disclosure database” which, in many instances, can store up to 72 hours of moment-by-moment data on any given patient until that patient is discharged. However, this is done at a much higher frequency than is normally stored within electronic health records.
H. The ability to communicate standardized alarm types and alarm violation types to the EHR in near real-time.
i. If a medical device generates an alarm, the alarm information and details are communicated to the EHR in time to support clinician life support efforts and critical care activities. Both text-based and audible alarm information should be communicated. For example, when a clinician or patient modifies device settings such as patient-controlled analgesics that are out of range and generates an alarm, the alarm and associated device details are communicated to the EHR.
Zaleski Comment 11: Alarm communication to EHR
If we expect episodic, regular, quasi-continuous, sampled waveforms, continuous waveforms to somehow be supported, then near real-time implies real-time to me. This presents an interesting quandary: if we are to communicate interventional information to the EHR, is it not implied that this information will, somehow, be used for interventional guidance? To me, this further implies a medical device, possibly requiring pre-market notification and substantial equivalent to existing monitoring systems and full-disclosure databases (Class-II regulatory implication).
I. The ability to set and communicate limits and safeguards for device settings from the EHR to a device.
i. Evidence-based guidelines or clinician preferences for device parameters or alarms may be communicated from the EHR or other systems to the device. For example, this would enable an infusion pump to be interrupted or paused based upon EHR information. Interrupts and pauses are not intended to be or imply closed loop control.
Zaleski Comment 12: Threshold and guardrail settings communicated to EHR
Presumably, we’re talking about communicating notifications to clinicians who would then intervene and stop or adjust the device, since we’re not talking about closed loop control.
J. The ability to wirelessly communicate point of care device information from the device to a device intermediary or EHR.
i. Wireless communication of high-acuity and inpatient medical device information may require specifications for wireless networking that supports the critical nature of this information and can co-exist with other medical devices and wireless applications.
Zaleski Comment 13: WiFi communication of medical device data
Clearly key: high quality of service, secure, available, reliable wireless infrastructure. This is the subject of an entirely new discussion, and one that I will be bringing up in the future. From iPhones™ to BlackBerrys™, clinicians of the future will be relying on mobile device technology and will expect them to support their clinical workflow in ways we have not yet even considered.
I present a concept for autonomic cardiac pacing as a method to augment existing physiological pacing for both ventricular assist devices (VAD) and heart transplantations. The following development represents a vision and reflects an area that has yet to be fully exploited in the field. Therefore, the analysis is meant to be a starting point for further study in this area. Furthermore, an automatic control system methodology for both heart rate and contractile force (stroke volume) of patients having either an artificial left ventricular assist device (LVAD) or who have experienced degenerative performance of the Sinoatrial node is suggested. The methodology is described both in terms of a device and associated operational framework, and is based on the use of the naturally-occurring hormones epinephrine, norepinephrine, and dopamine contained in the return blood flow through the superior vena cava. The quantities of these hormones measured in the blood stream are used to derive a proportional response in terms of contractile force and pacing of the Sinoatrial node. The method of control suggests features normally described using cyclic voltammetry, expert systems, and feedback to pacing an artificial assist device.
AVN, Atrioventricular Node
CO, Cardiac Output
FDA, Food and Drug Administration
LVAD, Left Ventricular Assist Device
NHLBI, National Heart, Lung, and Blood Institute
SAN, Sinoatrial Node
SV, Stroke Volume
TAH, Total Artificial Heart
TCI, Threshold Crossing Intervals
VAD, Ventricular Assist Device
Background on Controller Design
Artificial assist devices that exist today normally operate by controlling heart rate based on Proprioceptors–position of limbs & muscles during physical activity, or Baroreceptors–monitoring blood pressure in major arteries & veins. None to date operate on the basis of Chemoreceptors: monitoring changes in chemical makeup of the blood stream in direct response to epinephrine production by the adrenal medulla. Such changes are, however, more consistent with the operation of the human heart. For instance, heart rate varies not only according to mechanical movement of limbs, but also as the result of changes in emotion. Such changes manifest themselves as increases and decreases in sympathetic and parasympathetic hormones. Sympathetic hormones (epinephrine, norepinephrine) tend to increase stroke work and heart rate, whereas parasympathetic hormones (acetylcholine) tend to lower heart rate. These two hormones operate to control the resting rate of the heart and its changes as a result of higher-brain center changes (including production of hormone by the adrenal medulla). Benefits of achieving this capability include more naturally-behaving artificial hearts, or, in the case in which a human heart is merely being paced by an assist device, to control normal heart function in relation to changes in hormone production. Thus, the controller described herein provides an adjunct to existing controllers. Specific methods outlining artificial control mechanisms for affecting heart rate and contractility have not specifically been described in the literature, although related concepts have been suggested .
Autonomic Heart Regulation
Autonomic regulation of heart rate is controlled via one of the following specific systems within the body [2, 3, 4, 5, 6].
Proprioceptors – monitor general movement—position of limbs & muscles during physical activity
Chemoreceptors – monitor chemical changes in blood
Baroreceptors – monitor blood pressure in major arteries & veins
Chemical regulation (Hypoxia, Acidosis, Alkalosis)
Hormones (Catecholamines & thyroid)
Cations (balance of K+, Na+, Ca2+)
Current artificial heart assist devices operate using feedback from items (1) and (3) alone. Long-term (weeks to months) artificial heart assist devices popularly used for ventricular support, all under the auspices of NHLBI , include:
Abiomed extracorporeal, pneumatically-driven, pulsatile, left, right, or biventricular, introduced in 1988, FDA approved for in-hospital use for low output syndrome.
Thoratec extracorporeal, pneumatically-driven, pulsatile, left, right, or biventricular approved for in-hospital use for post cardiotomy low output and as a bridge to transplantation.
TCI (Heart rate) implantable , , pulsatile, pneumatically-driven, approved for in-hospital use as a bridge to transplantation. The electrically powered totally implanted configuration with a wearable power source, transcutaneous power lead and vent, is approved for in-hospital as well as out-of-hospital use for bridging to transplantation and is currently used under IDE in the randomized REMATCH trial.
Novacor  implantable electric, pulsatile, left ventricular (LV), wearable power source, transcutaneous power lead and vent, is approved for in-hospital and out-of-hospital use for bridge to transplantation.
Underlying Motivation for the Heart Control Method
The suggested device measures the chemical content of the blood flow past the sinoatrial node, processes the chemical content via cyclic voltammetry [11, 12, 13, 14] to determine the amount of epinephrine (EPI) and norepinephrine (NE), and uses this information to establish the resting and various pacing heart rates of a left ventricular assist device. Furthermore, researchers  focusing on advances in heart assist devices have identified characteristics desired for upcoming artificial organs. These include (1) miniaturization, (2) interfaces with nerves for automatic control, (3) control systems that are acceptable for both the living body and the embedded artificial organs, and (4) harmonization with the living body in various ways, including interfaces with higher brain centers and reduction of thrombus (and the associated foreign body rejection issues).
“the neuronal and hormonal control of the circulation, including the control of the heart, is mainly effectuated by the autonomic nervous system and its hormonal transmitters, the catecholamines. Autonomic control of the circulation primarily operates through the sympathetic system, though to a slight extent through parasympathetic signals to the heart. These have been lumped together, and there are basically three separate feedback mechanisms in this computational block. These are (1) feedback from the baroreceptor control system; (2) feedback from the peripheral chemoreceptors in the carotid and aortic bodies, and (3) feedback control of the circulator system caused by central nervous system ischemia, that is, ischemia of the vasomotor center in the brainstem. Another input that affects the autonomic nervous system is also included: The activation of the autonomic nervous system during exercise.” 
Methods for measuring serum epinephrine levels exist. WPI  suggests that sensitive, low noise carbon fiber (CF) and carbon disk (CD) electrodes can be employed in the electrochemical detection of catecholamines (e.g.: EPI, NE, Dopamine). As reported on their Web Site , the CF30-500 class of Carbon Fiber Disk Microelectrodes shows current output in Pico Amperes versus Dopamine concentration (nanograms/milliliter). In this analysis, WPI sites work by D. Yeomans and X. T. Wang of the University of Illinois. The analysis shows excellent linearity characteristics for CF filaments ranging in size from 10 to 30 microns to compounds with…detection limit[s] as low as 0.2 nanoMoles. Figure 1 illustrates this linear relationship .
WPI reports  that “longer CF electrodes…provide higher sensitivity and larger signal to noise ration. They are hence very suitable for in vitro amperometric and differential pulse voltammetry (DPV) in which…voltage scan rates are much lower.”
WPI further suggests  “extracellular recordings using CD electrodes (CD-30) in CA1 region of the hippocampus in an anesthetized rat shows ultra-low noise (<5 microVolts).” Voltage response in same region varied between +50 and –100 microVolts with response time less than 5 milliseconds. This is important from the real-time implantable device perspective as changes in responses need to be measured rapidly in order to correctly mimic actual behavior. A suggested bio-sensing device based on those implemented by WPI and others [22, 23] is illustrated in Figure 2.
While WPI stated  at the time of publication that there was no existing electrochemical method that exhibited selectivity among the members of the catecholamines or indolamines, others , reporting later, suggest that there may be ways to do this.
The Heart Conduction System
Now that the electrochemical mechanism has been suggested, let’s proceed with the automatic control methodology. Figure 3 indicates the hierarchy in which electrical signals and pacing are conducted throughout the human heart .
The Sinoatrial Node: Primary Heart Pacemaker
The SAN, the primary pacemaker for the heart, is located in the rear wall of the right atrium near the opening of the superior vena cava. The SAN has the fastest rhythm producing, on average, from 60-70 action potentials per minute. SAN pacing overrides all others. The action potential originated within the SAN travels through walls of the atria causing contraction. Internodal pathways connect the SAN to the atrioventricular node (AVN). The AVN is the located in the right posterior portion of interatrial septum. In the absence of SAN pacing, the AVN can take over, although the resting rate is slower (40-60 action potentials per minute). Next, the AV Bundle of His divides into both right and left bundle branches in the ventricular septum and is the only electrical connection between atria & ventricles. The Bundle of His is capable of generating from 30-40 action potentials per minute. The Purkinje fibers (4) are distributed throughout the ventricular myocardium and synchronize ventricular contraction. Ventricular muscles can generate from 20-30 action potentials per minute.
The Mechanics of Catecholamine Measurement
The SAN is that which is affected by stimuli such as adrenaline, exercise, drugs, etc. As discussed, chemoreceptors that detect changes in catecholamine levels (EPI, NE, Dopamine) translate into changes in pacing and contractility. Studies of environmental stress on epinephrine levels in humans show rather discernable relationships between EPI levels and work-related stress . Table 1 depicts the NE levels in male and female managers as a function of time in the workday .
Catecholamines: Regulate Heart Function
Catecholamines affect the sinoatrial cells as illustrated within the simplified cartoon of Figure 4 . The medullary cardiovascular control center in the brain contains both sympathetic and parasympathetic neurons that act as agonists to control contractility (ventricles), constriction (veins, arterioles), and control secretion of hormones to which the SAN responds. Carotid and aortic baroreceptors also respond to changes in blood pressure which provides feedback to the medullary control center that, in turn, affects contractility, pulse, and vasoconstriction.
The sympathetic and parasympathetic hormones EPI and acetylcholine are secreted from the adrenal medulla and have the effect of causing increased and decreased pulse and contractility, respectively . NE is a principal neurotransmitter in the sympathetic nervous system and is an a-adrenoceptor agonist , implying strong vasoconstrictor response, and, therefore, affects systolic and diastolic blood pressures as well as heart rate and contractility through b1-adrenoceptors . Metabolism of EPI and acetylcholine at the cellular level is illustrated in the diagram of Figure 5. Relationships between stroke volume and contractile force are well known in the literature . Thus, methods of measuring changes in catecholamine concentration must accommodate the combination of both pulsatile rate and contractility. The effects of sympathetic and parasympathetic nervous system on cardiovascular performance are known .
Changes in sympathetic and parasympathetic hormone affect the arteriolar smooth muscle fibers, the ventricular myocardium, and SAN to affect (as described previously) vasodilation, contractility, and heart rate, respectively. Studies have quantified the effects of changes in mean plasma catecholamine levels both in vivo and in vitro using techniques such as cyclic voltammetry and blood concentration measurement [35, 36]. Heart pacing set by SAN is regulated by antagonistic mechanisms: primarily through sympathetic innervations (release of NE, EPI – increases rate) and parasympathetic innervations (release of acetylcholine—lowers rate). These innervations act to regulate and control mean arterial pressure through heart rate, stroke volume, and constriction and dilation of arterioles, arteries, and veins. Measuring the amount of EPI, NE, and dopamine in vitro has been performed . One measurement approach is via cyclic voltammetry, details of which are available in the literature. Figure 6 depicts the author’s reconstruction of the relationship between carbon fiber peak anode current and Dopamine levels obtained via in vitro measurement using gold-tipped nanotube electrodes . NE and EPI are released from chromaffin cells. Discrimination between NE and EPI from the same cell has been reported to be possible using slow-scan cyclic voltammetry . Furthermore, the use of amperometry enables sub-millisecond release events to be measured . The mechanism for EPI and NE measurement via chromaffin cell ordinarily involves contacting the cell surface via microelectrode. Furthermore, placing microelectrodes at a distance exceeding several microns can result in significant signal loss .
Suggested Heart Controller Implementation
A methodology is suggested that provides an initial point-of-departure for refinement and iteration. The starting point depends upon the relationship between plasma concentrations of the sympathetic and parasympathetic hormones. Figure 7, Figure 8 and Figure 9 derive a rather simple relationship between HR and NE levels .
The relationship is somewhat misleading in that the implication is that NE is causative with respect to HR. This is only partially true, Figure 9 illustrates with respect to EPI. Again, the relationship provides only a partial picture. Heart rate is only one component affected. It is worthwhile to point out again that these relationships are not necessarily causative. The association between NE and EPI plasma concentrations is shown in Figure 9. Analysis indicates a linear relationship exists in EPI the ranges specified, and the correlation appears to be quite good, suggesting a relatively simple predictive model. While HR relationship to hormone level suggests a linear dependency within the specified NE range, another study , Figure 10 shows a more nonlinear relationship between cardiac output and EPI. This relationship is suggestive of an optimum level of CO change with respect to EPI concentration in dogs. While not conclusive, this relationship serves to illustrate the point that a nonlinear relationship can exist that must be represented in the modeling of plasma hormone levels and the effect on contractility and pulse.
Determine the resting pulse and cardiac output of a patient, where pulse (/min) x SV (liters) = CO (liters/min).
Measure the blood plasma EPI, NE, acetylcholine, and dopamine levels. This establishes the baseline state of the patient.
Measure the patient’s pulse and CO and draw blood samples associated with the plasma levels of the hormones during specific activities, including vigorous exercise, sleeping & awakening. This establishes the training set of inputs (i.e., hormone levels) and outputs (i.e., pulse, CO).
Build the expert system training model that establishes the inclusive range on hormonal input versus output parameters.
Calibrate the voltammetric sensor (invasive component) for measuring real-time anode current versus hormonal concentration. Anode current levels correlate to different voltage samples using cyclic voltammetry. Peak anode currents vary according to hormone level.
Thus, the specific level of each hormone level would be identified on the basis of the sample voltage value. The training mechanism involved relies on taking known inputs (e.g.: catecholamine levels) and measuring outputs, then using these values to develop a training matrix that establishes the transformation between the input and output (e.g., pulse, stroke volume). So, very crudely, this might be represented as follows:
wherein the xform(training) matrix is determined based on the input and output. Note that this is not a single matrix and not this simplistic in representation: an array of inputs and matching outputs will need to be determined that will translate into classes of transformation matrices. Of course, the viability of this approach would need to be determined. Furthermore, the outputs would provide only one component of input determinant to cardiac behavior. The effects of vasoconstriction, for example, must also be accommodated in terms of its effect on arterial pressure and loading.
In Vivo Operations:
Hormonal concentration derived value from cyclic voltammetry defines the input parameters (test parameters) used as input to the feed-forward expert system trained using the training set developed above.
Output pulse and CO in terms of pacing trigger voltage to SAN defines the derived pulse and, thus, the appropriate rate for patient heart function based on catecholamine levels.
This new pacing relationship between input hormonal levels and output pacing can be maintained for a specific patient within a processing chip associated or in proximity to a pacemaker unit. The trained relationship then establishes the expected behavior for a cardiovascular pacing or left ventricular assist device. The equation relating pacing to hormone level can be stored in a secure electronic patient record (for instance) for recall, updated training, or for use in data mining to compare and develop more complex relationships with those of other patients. While this methodology does indeed require validation and refinement, it defines a vision for possible implementation. The very nonlinear relationships among the input and output variables cannot be simply represented using one-dimensional mathematical relationships. Furthermore, mechanical and physical issues remain that will be challenging. For example, biofouling of in vivo electrodes must be overcome and represents a formidable technological challenge .
The preceding describes a rough and partially complete model based on lab research that is suggestive of baroceptor measurement of in vivo catecholamine levels. The motivation behind the approach is the lack of capability in current LVAD technologies that focus on this aspect of autonomic pacing. The methodology and concept would also apply in those cases in which patients may have damaged SANs.
As pointed out, detailed training issues related to generalization to any LVAD, validation of the range of catecholamine concentrations and impacts on pacing is that pulse and contractility are well behaved and do not pose a hazard to the patient, issues related to biofouling of sensors and calibration must be addressed. However, even before considering implementation, concept and technology proof-of-principle must be validated. This will require both human and non-human trials. Operationally, manufacturing and implementation challenges must be overcome.
 Y.E. Earm, Y. Shimoni, A.J. Spindler, “A Pace-Maker-Like Current In The Sheep Atrium And Its Modulation By Catecholamines,” J. Physiology (1983), 342, 589-590.
 University of South Australia—online learning environment–www.unisanet.unisa.edu.au/Information/12925info/Lecture%20Presentation%20-%20The%20Heart.ppt
 I. Kestin, “Control of Heart Rate,” Physiology, 1993, Issue 3, Article 3.
 NHLBI: “Expert Panel Review of the NHLBI Total Artificial Heart (TAH) Program: June 1998 – November 1999).
 CT Lewis et al., “The use of an implantable left ventricular assist device following irreversible ventricular fibrillation secondary to massive myocardial infarction,” European Journal of Cardio-Thoracic Surgery, Vol 4, 54-56, Copyright 1990 by European Association of Cardio-thoracic Surgery.
 Todd J. Cohen, “A Theoretical Right Atrial Pressure Feedback Heat Rate Control System to Restore Physiologic Control to the Rate-limited Heart,” Pacing and Clinical Electrophysiology 7 (4), 671-677, July 1984.
 R. Mark Wightman, “Probing Cellular Chemistry in Biological Systems with Microelectrodes,” Science 17 March 2006: Vol. 311 no. 5767, pp. 1570-1574.
 Jinwoo Park, et al., “Diamond microelectrodes for use in biological environments,” Journal of Electroanalytical Chemistry, Volume 583, Issue 1, 1 September 2005, pp. 56-68.
 D. Bhaskarab, CR Freed, “Changes in arterial blood pressure lead to baroreceptor-mediated changes in norepinephrine and 5-hydroxyindoleacetic acid in rat nucleus tractus solitarius,” Pharmacology And Experimental Therapeutics, Volume 245, Issue 1, pp 356-262, 04/01/1988.
 6th International Micromachine Symposium Special Lecture: “Artificial Heart Research by the Use of Micromachines.” Lecture by Sinichi Nitta, Vice President of Tohoku University and Professor of the Institute of Development for Aging and Cancer
 E. Naujokat, U. Kiencke, “Neuronal and hormonal cardiac control processes in a model of the human circulatory system,” International Journal of Bioelectromagnetism, 2000, Volume 2, Number 2.
 “Carbon Fiber and Carbon Disk Microelectrodes for Electrochemical Analysis and Electrophysiological Recording.” World Precision Instruments, March 1998. http://www.wpiinc.com/products/biosensing/carbon-elec/CFM_AppNotes.pdf
 Ulf Lundberg, “Catecholamines and Environmental Stress,” Summary prepared for the Allostatic Load notebook. Last revised September, 2003. Author sites L. Forsman, “Individual and group differences in psychophysiological responses to stress-with emphasis on sympathetic-adrenal medullary and pituitary-adrenal cortical responses.” Doctoral Dissertation, Department of Psychology, Stockholm University, 1983.
 Adapted from U. Lundberg, M. Frankenhauser, “Stress and workload of men and women in high ranking positions,” Journal of Occupational Health Psychology, 4, 142-151, 1999.
 G. Monreal, Staff, Cardiothoracic Surgery, The Ohio State University,: MadSci Network: General Biology “How and why does caffeine affect the pulse rate of a person?”, Michael Onken, Washington University, February 2000,
 Vicki R. Kee, “Hemodynamic Pharmacology of Intravenous Vasopressors,” Critical Care Nurse, Vol. 23, No. 4, August 2003.
 “A drug that binds a receptor of a cell and triggers a response by the cell…Often mimics the action of a naturally occurring substance.” Source: MedicineNet.com
 Jeff Isaacson, “Mammalian Physiology 1”, Lecture 11, UC SanDiego, Lecture 11, Fall 2006. Source Text: Human Physiology, 4th Edition (2006).
 University of California at Berkeley lectures on cardiovascular system and heart, 2004. http://mcb.berkeley.edu/courses/mcb136/topic/Muscle_Cardiovascular/SlideSet2/cardiac.pdf
 Christoph Dodt, Ulrike Breckling, Inge Derad, Horst Lorenz Fehm, Jan Born, “Plasma Epinephrine Concentrations of Healthy Humans Associated with Nighttime Sleep and Morning Arousal,“ Hypertension 1997; 30:71-76.
 Spencer E. Hochstetler and R. Mark Wightman, “Detection of Secretion with Electrochemical Methods,” Department of Chemistry, University of North Carolina at Chapel Hill.
 Spencer E. Hochstetler and R. Mark Wightman, “Detection of Secretion with Electrochemical Methods,” Department of Chemistry, University of North Carolina at Chapel Hill.
 Y-H Yun et al., “A nanotube composite microelectrode for monitoring dopamine levels using cyclic voltammetry and differential pulse voltammetry,” Prec. IMechE Vol. 220 Part N: J. Nanoengineering and Nanosystems, 2007.
 Spencer E. Hochstetler and Mark Wightman, “Detection of Secretion with Electrochemical Methods,” Department of Chemistry, University of North Carolina at Chapel Hill, pages 13-20. Biophysics Textbook On-Line, Victor Bloomfield, editor, submitted February 18, 1998.
 Ibid., page 13
 Ibid., page 22.
 Christoph Dodt et al., “Plasma Epinephrine and Norepinephrine Concentrations of Healthy Humans Associated with Nightime Sleep and Morning Arousal,” Hypertension. 1997;30:71-76.
 Michael B. Maron, “Dose-response relationship between plasma epinephrine concentration and alveolar liquid clearance in dogs,” J. Appl Physiol 85:1702-1707, 1998.
 Y-H Yun et al., “A nanotube composite microelectrode for monitoring dopamine levels using cyclic voltammetry and differential pulse voltammetry,” Proc. IMechE Vol. 220 Part N: J. Nanoengineering and Nanosystems, 2007.
[Heart Controller is one area of research interest. See my dissertation for my primary interest area in critical care.]
John R. Zaleski Book on Medical Device Data, Modeling & Simulation
This cutting-edge volume is the first book that provides practical guidance on the use of medical device data for bioinformatics modeling purposes. Professionals learn how to develop original methods for communicating with medical devices within healthcare enterprises and assisting with bedside clinical decision making. The book guides in the implementation and use of clinical decision support methods within the context of electronic health records in the hospital environment. Supported with over 100 illustrations, this all-in-one resource discusses key concepts in detail and then presents clear implementation examples to give professionals a complete understanding of how to use this knowledge in the field.
“Medical Device Data and Modeling” Content Overview:
Introduction to Physiological Modeling in Medicine: A Survey of Existing Methods, Approaches and Trends.
Simulation and Modeling Techniques.
Introduction to Automatic Control Systems Theory and Applications.
Physical System Modeling and State Representation.
Medical Device Data Measurement, Interoperability, Interfacing and Analysis.
Systems Modeling Example Applications.
Modeling Benefits, Cautions, and Future Work.
About John R. Zaleski, Ph.D., CPHIMS
At the time of publication, John R. Zaleski was vice president of clinical applications and chief technology officer for Nuvon, Inc. Previously, he was senior director & research department head (RDH) of Biomedical Informatics at Philips Research North America, and was product manager for the critical care product line and director of clinical research at Siemens Health Services USA. Dr. Zaleski holds a Ph.D. in systems engineering with a specialization in medical and biomedical systems from the University of Pennsylvania. Dr. Zaleski’s specialty areas are in medical device integration, modeling and simulation, and statistics.
I studied and developed models professionally early on in my career for use in the aerospace industry. I then had what could be identified as a life change and decided to go back to school. I attended the University of Pennsylvania where I studied Systems Engineering, but with the specific objective of focusing on medical applications of systems engineering and the use of prediction and modeling technologies in medical practice. My interests were in the areas of modeling or predicting future trajectories of states given a priori knowledge of the equations of state and past history. I developed a hypothesis and worked very closely with my medical advisor, C. William Hanson, MD and his colleague Albert Cheung, MD to develop an experiment surrounding prediction and modeling the post-operative pulmonary states of patients who had undergone coronary artery bypass grafting (CABG). The objective was to model the pulmonary function of these patients while they slowly recovered from the effects of anesthesia and analgesia. While the process has changed a little in terms of approaches and guidelines, patients who had undergone post-operative coronary bypass grafting (or CABG patients for short) were in a special class at the time: they were highly technologically dependent patients for whom complete pulmonary function was dependent on mechanical ventilation for breathing. As the effects of analgesia and anesthesia wore off, these patients would recover breathing function on their own. One of the challenges in the management of these patients was reducing the level of mechanical or external respiratory support in appropriate and direct proportion to their ability to sustain spontaneous breathing on their own. I became fascinated with the prospect in several ways, one of which being the automatic controls problem presented, in which the level of breathing support, under the supervision of the attending physician, was reduced appropriately and manually in direct proportion to the level of demonstrated respiratory strength. Decisions to decrease support and, ultimately, to remove the airway tube from the patient once they could support spontaneous breathing on their own, was an example of a process that was guided and managed by a clinical guideline in accord with institutional and field best practices. Yet, not every patient followed the guidelines exactly, and many external variables and influences were used in the makeup of the decision to reduce support. I developed a model of respiratory support reduction to anticipate and trigger, based upon meeting appropriate guideline-based thresholds, when and how a patient could be anticipated to respond over time.
Clinical Studies Using Medical Device Data–Data Derived from Patient Care Devices at the Patient Bedside
My study at the time evolved and included a complete assessment of the patient, from the moment they entered the operating room until they were extubated successfully—a process that on average could take a minimum of one to two full days. But, throughout the process, what struck and informed me was the use of models—both explicitly and implicitly. Understanding the purpose and function of the clinical process provided deep insight into the reason for having specific types of knowledge that could be used by the clinician to make decisions. Thus, clinical informatics could be improved and the reason for acquiring, processing, and presenting certain information in specific formats could be understood from a non-clinical perspective. This motivated me to look at the patient care management process from a different standpoint: instead of looking at technology and its use in the clinical environment, I strove to understand the clinical problems, the reasons for clinical decisions, and then evaluate the best way to bring the right information to bear to solve these specific clinical problems. The result is a subtle but I believe essential difference in the way at looking at treating patients. Technology exists to support the treatment of the patient. If there is no obvious or clear way to demonstrate its immediate benefit to the clinician and, therefore, the patient, then it is an impediment to both. Thus, I look at modeling in the same way: if we can demonstrate the basic reason and benefit for using specific techniques, then we can join forces with the clinician in providing the best care for the patient.
Zaleski’s Second Book on Medical Device Data and Clinical Informatics
Zaleski’s second book on biomedical device connectivity and clinical informatics… This one targeted on using data taken from biomedical devices for use in modeling and clinical decision making…
Available for order in October, 2010.
Topics related to clinical informatics covered in this text:
Book combines medical device connectivity, methods for analyzing medical device data, and the use of those data to assist in bedside clinical decision making, through the use of mathematical methods, predictive algorithms, and visualization: the essence of clinical informatics.
Dissertation Title: “Modeling Postoperative Respiratory State in Coronary Artery Bypass Graft Patients: A Method for Weaning Patients from Mechanical Ventilation”
“Physicians, nurses, and other health care workers are facing a problem: provide affordable, quality health care to patients while at the same time satisfy cost constraints imposed on them by insurance companies and government agencies. Cost-cutting measures in many industries, including health care, have resulted in down-sizing “solutions” which achieve their goal of reducing costs by eliminating personnel. This approach, however, can take a physical and psychological toll on those remaining care-providers involved in the daily activity of saving lives. Technology has made an attempt to come to the rescue of these individuals by enabling easy-access to data on patients within their care.”
“Much of this information, though, is in a a raw and unprocessed form, and is generally large in quantity. For the weary health-care provider, the effort involved in viewing and processing this information in real-time can be a deterrent to its use. Few places bombard the health-care provide with more real-time data than the Surgical Intensive Care Unit, or SICU. Patients arrive in the SICU from surgery, their lives dependent on the talents of the critical care staff and the proximity of life-sustaining technologies for survival. While it is important to maintain all of the patient’s physiological functions during the critical 24 hour period following surgery, two of the most vital are heart function and breathing. Whereas heart function is maintained through the careful administration of drugs to reduce the strain of pumping blood through the body, breathing is accomplished directly through the use of a mechanical ventilator which breathes for the patient until spontaneous respiratory function is regained.”
This Ph.D. dissertation documents the research and development of a real-time predictor of patient recovery and viability for weaning from postoperative mechanical ventilation.
Weaning from postoperative mechanical ventilation isa key process in surgical intensive care. According to the SCCM (Source: Society of Critical Care Medicine, 2006), ICU patients occupy only 10% of the inpatient beds, but account for almost 30% of the acute care hospital costs. A key aspect of care in ICUs relates to weaning from postoperative mechanical ventilation.
The following are links to my Ph.D. postoperative mechanical ventilation dissertation from 1996:
Clinical informatics relies on use of real-time data presented in various ways and using various analytics to assist the clinician in assessing status and trajectory of the patient.
A great deal of time on patient care in intensive care units is devoted to management of pulmonary and cardiovascular systems. This is illustrated along with key parameters or measurands in the following slide image:
Medical Device Data Assist in Patient Care Management.
A key objective in managing patients post operatively is maintaining their oxygenation, as shown in the following slide image:
Laboratory data are necessary to ensure proper blood chemistry, and also provide verification of patient viability to wean. An example of the results of three blood draws is illustrated in the following slide image:
Pulmonary measurements, such as tidal volume, are among many that can be taken from the mechanical ventilator through serial connectivity and provide a means of monitoring in real-time the respiratory performance of the patient over time. This is illustrated in the following slide image:
Data from Medical Devices Achieved through Connectivity Enable Collection of Clinically Relevant Data.
By collecting and overlaying all such data, measured and acquired in real time from the equipment at the bedside, it is possible to establish a contextual picture for patient state. This is a key aspect of live, bedside clinical decision making. An example of such data are illustrated in the following slide image:
From time to time I have been asked to provide explicit details on the mechanics and methods behind Haar wavelet transforms. The purpose of this post is to walk through two simple examples that demonstrate the use of the Haar transform relative to two one-dimensional signals (time signals). The details of the Haar basis and Haar wavelet transform are available elsewhere. The purpose here is to provide a simple example of how the Haar basis is computed using a simple tool such as an Excel spreadsheet.
4×4 Haar Wavelet Example Calculation
Let’s begin with the end product: the following is a 4×4 Haar matrix computed using Microsoft Excel:
The Haar Matrix, which we will denote Hn, is given as follows for the case of 4 data elements, denoted as Vn:
This computes to that shown in the figure above. The Haar wavelet coefficients are computed by first inverting the Haar matrix and multiplying by the output signal vector, Vn:
Haar matrix inverse is calculated using Excel and the result is shown in the following figure:
Haar MS Excel spreadsheet cell layout
Each cell in the Excel spreadsheet is computed using the following cell entry:
Where $B$2 corresponds the cell corresponding to the first row and column of the original Haar matrix and $E$5 corresponds to the last row and column of the original Haar matrix. The elements (1,1) at the end of the expression must include the components of each cell. So, in the example above, (1,1) represents the element in the first row and column of the matrix inverse. This is place in the first cell of the Excel spreadsheet corresponding to this element. The last row and column element would be (4,4). So, for example, the cells would be populated as such:
Time and signal vector chosen arbitrarily for this example is as follows:
A plot of this signal is shown in the next figure:
The wavelet coefficients are computed using the follow expression:
Excel spreadsheet calculation:
It is possible to cull coefficients on some basis, such as their magnitude with respect to the largest coefficient. We can arbitrarily impose a threshold with respect to the largest coefficient (-4.9497) and remove those coefficients (set to zero) that are at or below this magnitude. Suppose we set a threshold of 30%. The wavelet coefficients with 30% threshold imposed result in the removal of the second coefficient:
The signal can be recomputed using this culled set of coefficients. They are calculated using the following expression:
Excel spreadsheet calculation:
A plot of the signal with an overlay plot of the recreated signal using 30% threshold on the wavelet coefficients is displayed in the figure below. Note the comparison between the two signals, indicating some loss of fidelity owing to the removal of the wavelet coefficient. This is a crude representation of the effects of destructive compression on the reconstruction of signals:
8×8 Wavelet Calculation
The method can be extended easily to any dimension. Let us consider an application of the Haar wavelet transform to an 8×8 Haar matrix:
The inverse of this matrix is as follows:
The base signal is defined as follows:
A plot of this signal provides a convenient visual rendering of the data:
The wavelet coefficients are calculated in precisely the same way as the 4×4 example shown previously:
Finally, the imposition of signal thresholds (20% and 30%) is shown and the signal is reconstructed in the manner previously described, only extended to an 8×8 Haar matrix. The resulting plot with the wavelet threshold impositions are plotted as overlays in the following figure:
The AAMI Medical Device Alarms Summit was held October 4th & 5th in Herndon, VA at the Hyatt. There will be much published on the AAMI web site in this regard, and much in the way of out-briefs and collateral so I will leave the complete minutes and summary of activities and goings-on to those charged with doing so. However, I am compelled to focus on a few related themes that were referred to by several of the speakers and to which I voiced my opinion publicly during the meeting. I will do so relative to two specific speakers and provide the input that I shared at the conference during the public question and answer forums.
Medical Device Alarms Keynote Presentation
The keynote speaker on medical devices was given between 8:45 and 9:15 by George Blike, MD of the Dartmouth-Hitchcock Medical Center. Early in his presentation, Dr. Blike discussed that not much has changed since the 1999 Institute of Medicine Report To Err is Human was published. In that report, the IOM concluded, based on two studies, that between 44000 and 98000 Americans were killed each year due to preventable medical errors. While the options to diagnose and treat have increased measurably since that time, the complexity of the treatment process undermines the benefits. This is also the case of information complexity, in which the amount of information that is now available in electronic form exceeds the environmental space limitations surrounding the patient.
Medical device alarms: Redirecting attention to the important things
Medical device alarms, Dr. Blike continued, as a way to redirect attention, about redirecting attention “from something that is less important to something that is more important.”
However, uncertainty plays a large role in medicine, and the uncertainty about the meaning of alarms requires knowing much more than just whether a parameter is outside some norm or threshold. It is about knowing the context surrounding the patient. It is more than managing alarms as “nuisances” in the clinical space. While it is true that clinical staff can become “snow blind” to the continuing cacophony of alarms within the environment, the reason for reducing the alarms is not to reduce the cacophony, but to focus the alarms to redirect attention in a way that truly helps the patient.
Dr. Blike referenced Lucian Leape, MD: “Anesthesia is the only system in healtchare that begins to approach the vaunted six sigma level of perfection that other industries strive for.”
The management of a patient becomes more like a feedback control problem, in which the process blocks of Detect, Diagnose, Treat and Monitor are key to the closed loop system. In an environment where upwards of 40 parameters must be monitored over time (ICU), it becomes a multi-dimensional, multi-parameter feedback control system exercise. In environments such as the ICU where nursing:patient ratios may be 1:2 or 1:1, the care team must be able to detect problems, diagnose cause, treat, and then monitor. As part of this process, the medical device alarms identify deviations of the system state of the patient from the expected state.
Medical Device Alarms: Defining the Problem
The keynote address of Dr. Blike was followed by a panel titled “Defining the Problem: It’s More Than a Nuisance.” James Blum, MD of the University of Michigan Health System and Barbara Drew, RN, PhD of the University of California, San Francisco were the members of this panel. With no disrespect to Dr. Drew, who had an impressive and extremely interesting and informative presentation, I wish to focus on the comments of Dr. Blum because of the message he communicated in terms of systems integration and data management. I resonated most definitely with Dr. Blum’s presentation as it has been a rallying call of my own now for almost 20 years (indeed, a large reason for my entry into the field).
I took some photos of slides, and there are three that speak strongly to me. The first of these is the slide on physiologic monitors, shown below. The key point is that alarms, in general are not “smart”, and this is especially true of physiologic monitoring where, in which there is no “penalty for high sensitivity with low specificity,” and a general lack of data integration. However, the historic and retrospective data is not readily available nor can prospective analysis and projection be performed with the data. Moreover, these data are not integrated with the wealth of other information that provides context on the patient. While it is true that alarms that are of a critical nature (e.g.: ventricular tachycardia or asystole need no other context, there are situations which do (e.g.: O2 Saturation changes, respiratory function changes) and thus it is important to incorporate the context surrounding the patient into medical device alarms.
The second of these charts from Dr. Blum is one that resonates very strongly with me: Electronic Medical Records may be good charting instruments, but in terms of their clinical decision support capacity relative to real-time data, they are very mediocre. The reason I say this is because of the last two bullets on his slides: the data resolution can be limited (critical care charting ~ 15 minute intervals) and suffer from garbage-in:garbage:out. Because modeling of physiological systems often requires high fidelity and high resolution information, sparsely collected data will often miss crucial events that may be seconds in duration or less. For example, assessments of heart rate and respiratory rate variability can be quite important and predictive as to patient stability, as well as critical medical device alarms related to V-TACH or ASYSTOLE. The EMR is simply not equipped to capture these data trends based on the rate of data collection: the likelihood of missing these events all together is simply higher than capturing them at all.
The third and final of these slides is the following titled “Integration.” The essence of the problem with medical device alarms is that they are fairly “one-dimensional” in nature: they typically are associated with the patient care device (PCD) and its function (e.g.: physiologic monitor, infusion pump, mechanical ventilator, etc.) This does not mean that they are univariate, but rather they do not take into account the entire context of the patient and environs, as well as patient history, chemistry, etc. The Integration slide makes the point that multiple systems must be taken together–or fused–to provide an intelligent assessment of what is important versus that which is not.
Medical Device Alarms from a Systems Integration Perspective
The last of the three photos above defines to me the essence of effectively managing medical device alarms is through the application of systems engineering and systems integration disciplines. First, complete and unfettered integration of data, from medical devices through ancillary information systems (lab, PACS, EMR, etc.) are required. Next, laying out the use cases and scenarios related to the types of problems and conditions a patient can experience needs to be done in a holistic way that ignores vendor and device boundaries. This may involve integrating data, user interfaces, and creating methods that “feed” on the data available from multiple sources and assimilate it to produce integrated outputs. The display mechanisms are important but are secondary at this point: dashboards that allow singular access to information (much akin to avionics design) may be appropriate here. However, more important is the overall integration of information to provide predictive modeling, retrospective trending, and for evaluating scenarios on the fly. This takes a longitudinal look at the patient state in terms of everything about the patient. As Dr. Blum identified, there may be 40 parameters (give or take) that form the basic state of the patient.
Medical Device Alarms as a State Space Problem
When I began my career it was in the aerospace field and I focused on state space modeling of complex systems. This state space modeling often involved various forms of filtering, including Kalman and Batch Least Squares filters. The systems integration aspect of this modeling involved evaluating the trend or future state with respect to the current state based upon a system model representation of the entity being modeled. From this model, a projection could be made into the future. As was stated by Dr. Blike and others during the conference, medicine is one field where uncertainty plays a large part, it is perhaps naive to think that one could model the human being in ways that many in the aerospace industry do. However, 20+ years ago when I began my studies at the University of Pennsylvania and began my research into prediction and modeling at the University of Pennsylvania Medical Center, that is precisely what I was doing on a smaller scale with a specific class of patients. The subject of my dissertation was predicting the post-operative respiratory behavior of coronary artery bypass grafting patients, a unique class of patient in surgical intensive care units. Many of the concepts brought up during the Medical Device Alarms Summit resonated with me from the early days of my dissertation. One in particular was the idea of taking multi-source, multi-variate data and massaging into an assessment of outcome. As I did when I conducted my research, multi-source data from laboratory, patient record (history, demographics, etc.) were incorporated into the overall assessment of outcome. I know that as I followed patients from surgery through endotracheal extubation afterwards that I found many interesting relationships once all the data were laid out before me: relationships between re-warming time and patient’s anesthetic dosing; relationships time to begin breathing and time to extubate. The approach took into account the fact that there were uncertainties in the modeling. The objective was to establish a gross, coarse model of behavior by looking at the patient as a “black box.” Higher fidelity warranted more accurate modeling. However, approaching the patient as a system and taking into account all information is one approach I believe is the key to effective medical device alarms management.
Estimates by the U.S. Census Bureau expect the population of Americans aged 65 and older to increase by more than a factor of two between 2010 and 2050 . At the same time estimates of healthcare expenditure increases between 2007 and 2017 show an increase to nearly 20% of GDP in this period . These estimates were made prior to the recent financial crisis that began during the Fall of 2008. Further compounding this increasing demand and the concomitant increase in costs is the availability of allied healthcare professionals. Some studies  identify the likely decrease in the number of physicians entering any number of key specialty areas, including cardiology (20% decrease by 2020), geriatrics (35% of current demand met today), rheumatology (38 day average wait for a new appointment), and primary care (on the verge of collapse). Those of us who are baby boomers are on the leading edge of this demand and, in order to mitigate and minimize the cost impacts on our children, it is our challenge and responsibility to innovate and meet these challenges without passing along unnecessary burdens to our children and grandchildren.
Age-related ailments & managing age-related health issues
For most of us, aging means more frequent and severe afflictions. Taking care of our health by improving diet, exercising, and maintaining an otherwise active lifestyle is essential to ensure a high quality life. Even with increased vigilance chronic ailments can affect us later in life, brought on both by our genetics and consequentially due to the lifestyles we’ve led in our youths. Ailments such as dementia, coronary artery disease, Alzheimer’s, myocardial infarction, congestive heart failure, macular degeneration, osteoporosis, hypertension, chronic obstructive pulmonary disease, diabetes, and others take their toll. Managing chronic diseases is costly from a logistical perspective in terms of time and money. However, even more to the point, effective and quality oversight of patients with chronic ailments requires regular review, screening, and monitoring of patients. This is further complicated by the need to serve patients who lack the means or are physically incapable of leaving their homes for extended periods. Telehealth and remote monitoring are a means by which a case manager—an individual assigned to oversee the care of chronically ill patients within a home-health setting—can review patient information on a regular basis (for example, daily) and support both the patient and the primary care provider. Furthermore, Intensive care units and emergency departments are becoming more crowded. Individuals with insurance are going to EDs because they cannot find satisfaction in terms of prompt scheduling with their gatekeepers (family practitioners). The quantity of individuals with chronic ailments is on the rise (stroke, CHF, diabetes, COPD, etc.) This is in part due to the fact that people are living longer. At the same time the Medicare and SS systems will not be able to sustain the growth in population over age 65. This means that working individuals will increasingly bear the financial burden for us “boomers.” As a result of increased longevity and the fiscal challenges, the retirement age will increase.
Medical device data, and medical device integration to assist in aging and health issues
So, what do we do? Well, several things: first, technology in the form of remote data collection, reporting devices and software will become more prevalent: glucometers, BP cuffs, spirometers and associated software will be more readily available for direct communication with personalized electronic health records. If the purpose of a typical visit is to take BP and diabetic assessments, this can be handled most by collecting data at the point of care (home) and transmitting to the physician’s office for assessment. Such also applies to nursing and assisted living facilities. Next, the technical infrastructure required to transmit and store these data will be required. Paying for this infrastructure could come from a number of sources. One possibility: most everyone nowadays has access to cable television. Cable companies could offer devices that integrate with existing modems to collect and transmit data to the FP, together with complementary emails to next of kin (e.g. “Your mother’s BP as of 8:10 this morning was 145/89”). Other technologies that can be used to evaluate and monitor chronic ailments such as macular degeneration can further reduce costs by providing video cameras at point of care whereby opthalmologists can review retinal changes without requiring an elderly individual to be transported at expense and time to a hospital or office. In addition, support for remote consults via VoIP and video can be supported over the same network. This empowers the remote provider with the ability to interact with the patient All of these technologies are in use in remote pockets around the world today. But, they will become more prevalent. These technology implementations will reduce costs and provide for more personalized care in comfortable settings (homes). Of course, nothing takes the place of the tactile hands-on. But, for routine visits the above will be invaluable. In terms of the software technologies, personalized medicine will become the norm (eventually). Telehealth will be key. But, also, support for automated workflow in the acute care environment will need to be augmented. This means fully integrating all data into the enterprise HIS.
Remote patient monitoring requires medical device integration to facilitate health care management
The U.S. Department of Health and Human Services through its Office of the National Coordinator for Health Information Technology, published operational scenarios focused on providing key information to assist in harmonizing standards on the implementation, certification, and policy implications for robust remote patient monitoring . Included in this assessment are requirements on interacting with personalized health records and enterprise health information systems. The approaches to advancing remote monitoring include both seamless communication from medical devices at the point of care (i.e., in a patient’s home setting) and with a case manager and primary care provider both through electronic transfer, storage, and display of health information and remote video and audio interaction with patients in the same home health setting.
Remote health monitoring to empower the aging population
Technology is not the silver bullet, but those described above are key enablers for remote health monitoring. Of course, the use of technology carries with it the implication that sufficient underlying infrastructure exists. This is not always the case in remote areas of the country. Satellite, cable, and fiber optic technologies are fairly extensive within the continental United States, but pockets and regions exist in which this is not the case. Therefore, a combined effort to extend the communications infrastructure must continue together with a unified effort to standardize and train and “in-service” individual care providers on these technologies must occur. One of the best mechanisms for enabling this is through the local hospitals and their satellite clinics.
So, how long do we have? Well, the sooner the better. Successful telehealth and remote monitoring programs exist throughout the United States and worldwide today. We should ensure that our elected representatives direct healthcare expenditures towards several specific areas to promote growth and alignment to meet the objectives of remote monitoring. These include continuing alignment on electronic personalized health records, expansion of our underlying communications infrastructure, and promoting common standards of communication among these records so that, regardless of location, a patient can communicate his or her information to any physician and allied health professional within the country. In summary: common storage, homogeneous communication, standardized formats.
 Source: Population Division, U.S. Census Bureau, August 14th, 2008; Table 12: “Projections of the population by Age and Sex for the United States: 2010 to 2050 (NP2008-T12)”
 Cinda Becker, “Slow: Budget Danger Ahead,” Modern Healthcare, March 3rd 2008.
This work combines much of the experience learned in medical device interoperability and clinical informatics I have gained over the course of the past 20+ years.
I have leveraged work from my Ph.D. and experience in product management of critical care. The device connectivity experience and lessons learned are documented in my first book. Text is available at Amazon.
“This cutting-edge volume is the first book that provides practical guidance on the use of medical device data for bioinformatics modeling purposes. Professionals learn how to develop original methods for communicating with medical devices within healthcare enterprises and assisting with bed-side clinical decision making…”