Medical knowledge coding
Preliminary experimentationsIn 1986, CAI (Computer Assisted Instruction) Club of school of medicine, University of Tsukuba started self learning group on Artificial Intelligence. Activities are chiefly reading elementary readings and some text book on LISP or Prolog.
In the same year, very small trials to code the medical knowledge delived from several famous medical text book into Lisp based Expert shell were conducted.
In Science Information Processing Center(SIPC), University of Tsukuba, an Lisp backend processor " FACOM alpha" manufactured by Fujitsu Co.Ltd. was operated. This system served as backend processor for Fujitsu
M360AP (IBM MVS/XA compatible Mainframe) to enhance calculation of UTILISP (University of Tokyo, Interactive LISt Processor). FACOM alpha was developed in Fujitsu Laboratory and was commercially shipped in 1985. While it was used not only for research purpose but also for controlling furnace temperature in
ironworks, sold volume was only 30 set.
"E-SHELL" (an expert system shell with forward and backward inference engines manufactured by Fujitsu Co.) was installed in M360AP for experimental purpose.
In first, medical knowledge coding was tried to code on this E-shell system and success fully diagnosed as if interview session in the consultation room by physician in several common diseases.
In 1988, infrastructure of the trial was moved to Prolog-KABA which is possibly most popular implementation of Prolog in Japan. This was the influence of expected replacement of M360AP system and the trend in AI studies in Japan in those days.
With these experiments, the difficulty in clarifying medical inference logic of physician became great problem to implement medical know ledges. As a result, this project was entered into long sleep so as Mr. Nagase to acquire logic of medical decision as expert.
Earlier Studies
Search and testBetween 1990 and 1996, search for earlier studies on medical artificial intelligence had been conducted slowly, including obtaining earlier implementations and to test the function and limitations of such system.
Intermediate goal settingIn 1998, intermediate goal was settled on introducing and implement Clinical Decision Support System on modern Computer Physician Order Entry System. Objective goals included (1) integrated patient data repository, (2) universal data access scheme to this repository from CPOE code. (3) effective migration strategy.
>Between 1998 and 2000, several site visit to renowned medical information system were made to acquire knowledge and skills to system migration and operations.