From the Boidem - 
an occasional column on computers and information technologies in everyday life

December 28, 2008*: Virtual worlds, and real health.

  This month's Boidem column is considerably different than all previous columns, because it is primarily a "guest column". The main text was written by Dr. Lawrence Fagan, a friend from almost before I can remember with whom I reestablished contact in the past few years. I've tried to react and comment, and in general "Boidemize" this main text through the links. Readers are invited to glimpse into how this column came about from both of our perspectives.

Web-based applications are becoming increasingly popular for supporting the information handling requirements of medical care. (This calls for some detective work.) One such application area that is rapidly expanding is the area of personal health records - web-based patient access to their medical information. Now that electronic medical records are being installed in many physicians' offices, health systems are making a patient's medical records accessible to him or her via the web. The patient logs into a web application to view pages that contain the current list of medications, laboratory test results, lists of past visits to the doctor, and probably most importantly, upcoming visits (and their location), laboratory tests (and any special preparation for the test), and the ability to order medication renewals. I use the Palo Alto Medical Foundation Online patient information tool developed by EPIC demonstrated here.

The information available to the patient via these online records is extensive, and this is information, such as a comprehensive list of prescribed medications, that normally is only available to healthcare providers. Additional advantages of this interactive system are the ability to renew medicines in a few clicks, or to ask the doctor to review your symptoms through a structured secure email message. This works well if all your medical care takes place in one clinical center. In the United States, however, patients often change medical centers at the annual open enrollment period. With each change, their records are electronically stored in disparate medical databases that are disconnected from each other. Ongoing work is taking place to create and use standards for communication between facilities, and more importantly for creating regional information pooling facilities that contain the most important information aggregated from each clinical center.

The problem of getting your personal medical information is most difficult in the case of illness during a trip or during a disaster such as the recent hurricanes in the U.S. During the aftermath of Hurricane Katrina, for example, patients and their doctors were separated by hundreds of miles. Patients with printouts of their medications (or perhaps, in the not very distant future, a secure USB drive with their medical information, including x-ray and medical scans), would have the key data with them, and wouldn't need an intact medical information infrastructure in order to access their patient data. After Hurricane Katrina, medical teams reconstructed the medication records of patients by merging the databases of local pharmacies and making this data available to physicians who logged in to a secure web site. But this is, of course, workable without having to wait for a disaster.

Google and Microsoft and a number of institutions are working on linking the information from all these electronic repositories to provide a comprehensive patient oriented record drawn from multiple medical care sites that can be especially useful in the care of chronic illnesses, such as diabetes or cancer. These patient-specific descriptions of the medical situation can then be linked to other internet resources such as: secure email to medical providers, medical research results, listings of treatment possibilities, and access to virtual groups of patients with similar problems. This allows for new possibilities for patient education in the context of the clinical situation provided by these patient health records. For example, a teenage asthma patient, or an elderly diabetic patient may need different information about their disease than a 40 year old with the same problem. Of particular interest is how to really take advantage of computer graphics to provide educational assistance.

This year, for the first time, a mainstream medical computing conference, the American Medical Informatics Association (a link on the upper left part of the page leads to a PDF listing of the conference program) showcased the use of virtual worlds to assist in patient education. The process of a patient leaving the hospital (called a discharge) is extremely important. Discharge planning includes a discussion between patient and nurse about all of the medicines being taken (especially the ones that are started or changed during the hospital visit). Reviewing this information with a patient can take longer than an hour of a busy nurse's time. At the conference researchers demonstrated a virtual nurse avatar that reviews this information (drug dose, time of the dose, etc.) with the patient. The patient uses a touch screen via which he or she verifies that they understand the information. The program takes virtual breaks and engages in social conversation to avoid tiring the patient. The document of instructions shown on the screen is coordinated with a hardcopy document that is presented to the patient upon being discharged. At the end of this virtual session, a human nurse intervenes to verify that all the patient's questions have been answered and to provide the true human interaction missing in the virtual interaction. Even so, patients surveyed about the use of avatars for presenting this key information noted that they overwhelming preferred the virtual nurse to human providers. Researchers are trying to understand the reasons for this. One possible explanation is that in the virtual session patients feel that they have the full attention of the virtual nurse. This calls to mind often cited research results that patients preferred to give drug, alcohol and sexual medical histories to a computer instead of a human provider. The assumed rationale for this result is that patients realize that the computer will not draw value judgments during the interview. Much has been published about this. Abstracts to some of the literature on this topic can be found here, here and here.

Another virtual teaching situation described at the conference was the use of 3D worlds for disaster planning - an influenza epidemic for example. With tools available today, a virtual hospital can be set up, and individual participants such as hospital employees or first responders can take on various roles - performing triage for victims, providing quarantine areas, examining patients, dispensing medications, and more. Significant computational and time resources are required in order to set up an environment, including: 3D models of the buildings and equipment, storyboards/scripts for the scenario, and programming animated movements that make sense for the actors following the scripts. Because the same hospital simulation - the 3D descriptions of the hospital building and its contents - can be used for other purposes such as orienting employees to the physical layout of the medical center, some of the development costs for a program such as this can be shared, and the basic scenarios reused. The demonstration I viewed took place within Second Life, but custom virtual world development systems also exist. We are still very early in these PC-based 3D programs for medicine, although some specialized areas, such as anesthesia simulators with realistic manikins, are very far advanced (a list of these is here). The 3D environment must be realistic enough that users suspend disbelief and become immersed in the simulation. This certainly happens in the anesthesia simulations, where it is not unusual for the clinician in training to express real concern when the simulated patient encounters a problem. In order for the observer to really believe the situation, the details have to be passably realistic. Special items such as hospital masks or decontamination equipment need to be designed in 3D modeling software and the acceptable actions, like putting on or taking off the mask, need to be specified for each element of the simulation. Pressing F5 to put on a mask has no relation to the action in the real world of finding the right size mask, removing it from the packaging, and placing it over your face. But if the images are realistic, and the flow of activities is meaningful, then pressing F5 or similar meaningless actions can be overlooked. Too many computerish steps and it is like being in a movie theater and losing the sense of being inside the action, looking around and thinking how much longer is this movie going to run.

This leads to my major complaint about the timeline of the simulations. Some actions that happen instantaneously with the click of the mouse button in the virtual world can take a great deal of time in the real world. Within Second Life a decontamination tent might instantly appear in the middle of the disaster area, but in real life chances are good that the tent is locked away in a closet, and nobody can find the right person with the key. And, of course, then the staff has to drag the tent outside and physically set it up. In Second Life or other simulation environments, we ordinarily don't see the actors sweating, or needing to stop to drink some water. Making a simulation of this sort realistic, particularly in the temporal dimension, presents a significant challenge. When a simulation is successful and convincing, participants actually do suspend their disbelief, and plunge into the action, remaining with it for longer periods of time than passive media demonstrations. Done well, this can be quite an intense experience.

After more than 30 years of work in computers and medicine, we are beginning to see the use of sophisticated programs that extend beyond medical professionals to patients and students. These programs take advantage of web access, sophisticated user interfaces, and years of work on standards for transmission of data. In the not too distant future, many patients' medical care will be tied to electronic media connecting health records with active sensors in the home to transmit updated blood pressures and other ongoing measurements. Meanwhile, we see advanced graphical interfaces, such as 3D virtual reality being used to improve medical care. In these two quite different examples, individual patients, hospital employees, and first responders are using experimental virtual environments to enhance both the transfer of information, and the training/learning experience. With all its possible drawbacks and difficulties, virtual world simulations may be the preferable method for this sort of training. When human models with mock injuries are used, simulating the physiological reactions of the body to injury is quite difficult, and definitely stretches our imaginations. Doctors certainly learn from the physiological reactions to real injuries - but inflicting injury for the sake of training goes against the Hippocratic Oath.

That's it for this edition. Reactions and suggestions can be sent to:

Jay Hurvitz

back to the Boidem Contents Page

Return to Luftmentsh