Generally when one looks for some part of human life to computerize, it is useful to limit the area being covered to the bare minimum in that endeavor so that you can slowly scale up the computer coverage from a small working base to a more capable system that covers the day-to-day aspects. This is difficult when the subject area is medicine, because almost any interesting subset of medicine is as complex as the set of medicine as a whole. So you find that bringing in a computer to help a small area quickly means that you are using the computer to handle a huge area. It is as if the only type of problems are toy problems, and life-critical practical systems. This also means that it is usually easier to teach someone who already knows medicine how to get the computer to work for them, than to take someone who is a computer specialist and try to explain to them all the intricate interactions which come about because of a particular medical specialty. VistA, by the nature of its unique history, did that far more than a lot of other systems of comparable size.
VistA started as an effort within the VA hospitals to bring in computers to enhance the provision of care to a largely stable population. The explosion of new veterans after the Vietnam war meant the methods of healthcare delivery developed after World War II simply could not keep up with the demands.
In non-veteran healthcare systems, a constantly changing patient population only come to the doctors or hospitals when an acute care need arises. The medical situation for the non-veterans are at the end of a disease process, when preventative care is no longer very effective, and some acute distress forces them to seek care. The nature of their visits means that they are more willing to spend money to get "fixed", but it also means that the only time the care system receives money is during the crisis. When one reviews the impact on the computer systems, they primarily track billing expenses, and itemize care given, with an underlying assumption that there is no need to keep long term records because when that person is seen again, their needs will be drastically different than this time, because it will be some other acute distress as this one will go away after they leave the hospital or clinic.
In contrast to this scenario, within the VA, the large number of veterans who are seen are seeking care because of an incident that occured when they were in active military service, and this service-connected problem has a continuing impact on their lives. Also, because the patients have already "invested" in their care by serving in the armed forces, the cost of their care is covered by allocations from the U.S. Congress, which usually come in the form of a fixed amount of money per patient per year. The VA, as stewards of tax money, wisely takes a longer view of healthcare, and deems preventative care as a way of dealing with disease processes before they reach a critical acute incident. Since there is a long term commitment to these particular veteran patients, the computer systems also reflect the chronic nature of this reality. Once a patient record is established, it will be available for years to guide healthcare providers. Long term trends as well as cumulative analysis is possible due to the wealth of data kept about a patient. Security and privacy of records is built into the design of the computer system at a deeper level, because there is more information to safeguard. Since many VA facilities are affiliated with a medical school, computer decision support is part of VistA to remind new practitioners of interactions between the various drugs, tests and pre-existing conditions. Since students and residents by the very nature of their activities are transient, checks and balances are built in to allow the established providers to review their work, and reflect the local standards of practice within the customized behavior of the computer and healthcare system.
Since VistA is tracking both healthy and ill patients, it has to track a wider range of human behaviour. The system has to model why particular activities are in the patient record, rather than just recording what occurred in a limited acute incident. The VA has been at the forefront of standardizing medical vocabularies in numerous areas, such as the National Drug File, LOINC coding for lab tests, the HL7 data transmission protocol, describing diagnoses and procedures with specialized nomenclature and coding systems such as the UMLS, ICD-9, CPT, DSM-IV or the SNOMED CT systems. Each of these systems has a lot in common with modern ontologies used in knowledge engineering and the semantic web. The actual syntax or methodology may take a different form than a typical expert system, but many of the basic operations of classification using generalization and specialization, analysis of processes of data collection and deductive processes are in common. And as has been said before, the subset of the world that has to be considered to be described or axiomatized is very broad. The area of overlap between ontologizing the world at large and ontologizing the medical world is large enough that almost all of the same issues come up in both efforts.
An explicit ontology of the medical aspects of VistA would involve developing methods of exposing a lot of the medical knowledge in the system in a form that external reasoning systems could then be leveraged to enhance the support of medical practice already provided by VistA.
No comments:
Post a Comment