"Emergency Medical Hologram" May Be Less Futuristic Than You Think

This weekend I was playing around with Prolog- a logic programming language (developed for natural language processing) often used in artificial intelligence programs.

Now, my program wasn't very complicated. It was designed to help me choose which of three pair of shoes I wanted to buy.

The top portion of the program was dedicated to entering information about the shoes: their cost, what material they were made of, arch support, ease of washing, ect. The second portion was a set of rules and superrules.

The rules taught the program how to categorize the random bits of information. For instance- I taught the program that ecologically responsible shoes were not made of leather. Practical shoes are not too expensive, and can be washed. Comfortable shoes had good arch support. Cute shoes are blue or yellow, not black.

The superrules used the rules and detailed information to make value judgements. It learned to tell me that acceptable shoes met the criteria for ecologically responsibility and practicality. Optimal shoes meet the conditions of acceptable shoes and are also not too expensive (and possibly cute)!

That may seem like a lot of work for picking out your shoes. But just think of the enormous machines that filled the basements of universities- the original computers- which did with thousands of dollars and thousands of man hours what would take a calculator of the day no trouble to reproduce.

The power comes with what you can do with more complex questions. What if, for example, I was not trying to match shoes to my quirky wants and needs- but trying to match the bizarre symptoms of a patient to a disease?

Turns out this has already been done, using another logic programming language called Lisp. Mycin was developed in the 1970s to help doctors diagnose and treat rare blood diseases. According to wikipedia, Mycin's inference engine, loaded with information not unlike my standards for selecting shoes (though much more complex), accepted yes/no input from a long series of text questions, and then listed the likely bacterial interloper, complete with a list of other likely candidates ranked in order of its confidence in the diagnosis. Mycin outperformed general practice physicians in diagnosing these rare blood diseases, and held up pretty well against specialists in the field.

Mycin was the predecessor to CADUCEUS. CADUCEUS's amition was to diagnose more than blood diseases- extending to about 1,000 different internal medicine condtions.

So the "Emergency Medical Hologram" seen in Star Trek: Voyager wasn't as far out there as he seemed. We don't have any holograms that can wield a scalpel- but we do have artificial intelligence inference engines capable of flipping through every condition and symptom known to man, and making a guess at diagnosis and treatment.

I wouldn't want to give up my health insurance just yet- trained professionals will always be necessary to assess the patient and enter the symptoms, to review the findings, and administer treatment. But imagine what a difference such a system could make to a third world clinic staffed by one overworked doctor. Imagine how a system like CADUCEUS could help stumped doctors diagnose a condition with irregular symptoms. Imagine colonists far away from Earth, their physician killed, and decades left before the replacement will arrive.

The possibilities are endless...and they all begin with one little line of code.