A voice-activated “heads up” 3-D audio and video display for the anesthesiologist of the future will provide all monitoring data and other patient-care information.
I have been asked to provide a “pie in the sky” vision of what our perioperative information management systems will look like in the not too distant future. Obviously, if the current system of managing information perioperatively was perfect, no one would have asked me to pontificate on this subject. In fact, there are a great many deficiencies in this aspect of anesthesia practice (and medical practice more generally). I will first specifically discuss a few of the most egregious problems. The remainder of our current shortcomings can be discerned by inference from the subsequent discussion of my vision of the system of the future. Due to the necessity for brevity, many interesting details and supportive data have been omitted.
Problems with the Current State of Perioperative Information Management
Throughout the perioperative period, but especially intraoperatively, we are given too much data, yet rarely enough information. One must distinguish data (“the systolic blood pressure is 95”) from information (“the blood pressure is low for this patient compared with their baseline systolic blood pressure of 155 torr”) from knowledge (“this degree of hypotension is likely causing inadequate coronary oxygen deliveryÑcheck for evidence of myocardial ischemia”). All too often it takes way too much effort to get the right information. What data are available are often unreliable or uninterpretable. How many times have we reviewed a manually created anesthesia record from a patient’s previous surgery and been unable to decipher the critical information. Sometimes we get patently wrong information (e.g., the patient tells you when you ask, “I don’t know why the chart says I am allergic to Pentothal, I am really allergic to penicillin”) or we get correct information but not in a timely manner. How often in your institution does critical data finally arrive, long after it would have been most useful. Finally, data are frequently presented in a way that is less than optimal. Regardless of how the perioperative information management system of the future evolves, it must address several key clinician needs. First, what every clinician really wants in terms of clinical information is just what is needed (no more, no less), just when it is needed. This information must be presented in optimal order and format for each specific situation at hand. The information must be accurate, reliable, and sufficient to support (preferably augmenting human knowledge and capabilities) the decisions that must be made at the moment.
The Perioperative Knowledge Database
The core attribute of any perioperative information system must be a comprehensive integrated medical database (or, in light of its anticipated intelligence, “knowledge base”). This information repository must contain, in a highly standardized format, all essential medical information about every perioperative patient. By necessity, the knowledge base will be distributed through thousands of computers around the globe. The key will be to network all of the pieces of the whole so that a clinician can access absolutely everything relevant about a single patient from any single location no matter how remote.
It is important that both process and outcome data are incorporated. At each site of care, the perioperative information system must keep track of (and archive) what is done to each patient (both diagnostically and therapeutically), how, when, and why it is accomplished, and how the patient responds. Patient education activities, the nature and substance of doctor-patient communications, and patient satisfaction must be incorporated. The system will rely heavily on artificial intelligence to integrate individual data elements, to infer relationships, and to identify data deficiencies (data that are missing or probably wrong).
Importantly, the knowledge base will provide not only individual patient data but also relevant population data (e.g., complete information about all of the patients in the United States with coronary disease and COPD who have had a carotid endarterectomy). The individual practitioner will thus be able to do knowledge searches about patient populations similar to his/her own to relate, for example, preoperative characteristics and anticipated anesthesia/surgical procedures to population patient outcomes. The anesthesiologist will also be able, at the same time from the same network interface, to review relevant evidence-based practice guidelines and all applicable medical knowledge. These different knowledge bases will be integrated seamlessly.
Experience with electronic anesthesia record keepers to date has reinforced the computer science maxim “Garbage In Ñ> Garbage Out”. The perioperative system of the future will assure that individual data elements are of the highest possible quality and reliability. In fact, the system will assign levels of reliability and uncertainty to each data element and this information will be used intelligently in any inferences drawn by imbedded expert systems.
Obviously, such a comprehensive repository of information about individuals must be carefully guarded from misuse. All personal information will be highly encrypted and password protected, as will transmissions from computer to computer. Each data transaction will, in fact, only be initiated after explicit electronic approval of the involved patient. First, definitive patient identification will occur with DNA sampling. Then, patient permission will be obtained via fingerprint, retinal or voice scan. Clinicians will only gain access to these data after similar security procedures.
Managing Information in the Preoperative Period
The clinician seeing the patient preoperatively will have available all of the information necessary for care planning and patient education. The patient’s medical history, medication history, prior encounters with anesthesia care, laboratory and imaging studies, and preoperative surgical information will all be electronically accessible, instantly, and in a usable format. The system will contain a complete “audit trail” of all interactions between the patient and the healthcare system including the where, when, who, what, and why of each medical test, prescription, procedure, and clinician visit.
Discussions with primary care physicians, specialist consultants, and colleagues will all occur instantly through internet videoconferencing. For patients who can not be seen in person, the preoperative evaluation will be similarly conducted via computer video. Upon completion, the preoperative evaluation, including any new data generated, will be automatically added to the patient’s electronic medical record. Entry of annotative data (e.g., anesthetic plan) will be effortless, likely by natural language voice recognition.
When the patient arrives on the day of surgery, a personal information management (PIM) unit will be attached to their wrist. The PIM will continuously monitor and record the patient’s heart rate, temperature, oxygen saturation, and other physiological variables. The PIM will not only provide a continuous indication of the patient’s location but will also orchestrate the transfer of all patient-relevant information throughout the facility. For example, when the patient enters the operating room, the PIM will transfer all patient physiological vital signs to the displays in the OR, and will notify the OR scheduling/management system and the Family Information Update System of OR entry. Additionally, the surgeons will automatically be paged.
Managing Information Intraoperatively
The anesthesiologist in the OR will have electronic access to all relevant preoperative information as a part of the patient’s complete medical record, presented upon request in an optimized prioritized format. The patient will no longer need to be moved physically from a guerney to the OR bedÑthe two will be integrated into a single all-purpose transport system. All physiological sensors will transmit their signals from low-profile transducers on the patient via wireless channels to a central processor.
New sensor technologies will measure directly critical system variables to enhance monitoring and decision making. The anesthesiologist will have available real-time continuous plasma drug levels (e.g., propofol, remifentanil, etc.) and levels of relevant endogenous substances (e.g., Glu, K, Na, pH, O2, CO2, Hgb, troponin, etc.). A reliable continuous measure of “level of consciousness” based on brain and spinal cord function will be available. New sensor technologies will provide real-time continuous non-invasive measure of organ function at the whole organ (e.g., cardiac output), component (e.g., left ventricular end-diastolic volume), and intracellular (e.g., myocyte [ATP], phosphorylation state of stress activated protein kinase) levels. Clinicians will be particularly interested in monitoring non-invasively energy, electrical state (transmembrane DC potential) and chemical ion (e.g., intracellular calcium or potassium) levels in specific organs (cerebral, cardiac, hepatic, renal, and peripheral tissue). Non-invasive imaging technologies will have advanced tremendously and allow real-time use intraoperatively to guide surgical progress. Surgical information systems will be fully integrated with anesthesia information systems to enhance clinical communication and team decision making.
With ever increasing amounts of data available to monitor, organization and visualization strategies gain in importance. Computer-based decision support and intelligent oversight will be ubiquitous. Record keeping will be truly automated and, when user input is required, it will be effortless (generally using natural language speech recognition via personal microphones).
User interface design will have advanced tremendously. These intelligent control and display systems will be based on years of extensive research and testing. The anesthesiologist will wear a lightweight voice-activated heads-up 3D audio and video display. Although most information will be displayed visually and audibly, other interface modalities (haptic, tactile, and olfactory) will be employed where appropriate. Highly effective decision support systems will utilize artificial intelligence technologies (expert systems, neural networks, genetic algorithms, etc.) to integrate baseline and trended patient information with predictive models based on population data and evidence-based guidelines. Most information will be presented in a “mixed reality” with video and computer-generated images over-layed on the actual clinical view.
Haptic interfaces will allow direct control of robotic systems that assist in the conduct of interventional procedures such as direct laryngoscopy. The selective use of automated systems, such as closed-loop control of intravenous drugs (anesthetics, analgesics, neuromuscular blockers, and vasoactive drugs), will be facilitated by ecological interface design and voice activated supervisory control.
In the event of an emergency, the information management system will automatically provide decision support (broaden prioritized differential diagnosis, appropriate treatment algorithms, etc.). The system will also assume control of routine functions to allow the clinician to focus on patient care, and will collect, like a cockpit flight recorder, all relevant physiological, management, and voice data.
Of course, these complex technologies will require user training and certification. This will occur initially in realistic OR simulations. But training will be continuous during actual clinical performance using intelligent automated tutoring and mentor software.
Managing Information in the Postoperative Period
When surgery is complete, there will be a seamless transfer of information to postoperative care team(s). All patient data will be automatically incorporated into the patient’s permanent medical record and all relevant population database(s). In addition, the outcome of each patient’s anesthetic will be automatically compared with current evidence-based guidelines and these will be incrementally adjusted as appropriate. Where appropriate, patient monitoring and surveillance will continue in the patient’s home using wireless communication link. All patients will receive videoconference-based postoperative visits. If the patient has any undesirable outcome(s), these will be automatically input into QA database and all relevant providers will be notified.
Although the preceding may read like a science fiction novel and much of it may never come to pass, my vision provides the outline of a goal to which both clinicians and manufacturers can collaboratively aspire. It is going to take some hard work (and a bit of time) but our patients will certainly be appreciative.
The author acknowledges the contributions of Drs. J. S. Gravenstein and D. M. Roth to this paper. Preparation of this paper was supported in part by the National Patient Safety Foundation and by NIH subcontract HL64590 (Westenskow). Some of the research upon which this paper is based was supported by grants from the Anesthesia Patient Safety Foundation and HL64590.