Medical artificial intelligence and zoology motivation knowledge inscribe ②

Medical Interview Preparation - Duties of a Doctor

[medical artificial intelligence and zoology motivation knowledge inscribe] : Intelligent history date from

Software of medical of family approach force (Shenzhen) limited company Likewei

One, weichi to rich from numeric computation train of thought

Will look at present, 3 of the computer main fields are existing it seems that layer class goes forward one by one relation: General affairs of numeric computation → handles → artificial intelligence. Actually, inchoate computer natural resources is very scarce be short of, can consider only on explore major event or most advanced issue; Besides multifarious numeric computation, traditional intelligent issue is worn by a lot of people‘s historic attention. Accordingly the computer was born 1946 before long, graph forest put forward ‘ artificial intelligence ‘ concept. After, had the have more than needed on certain level when computer natural resources, of vulgarization ‘ general affairs is handled ‘ just be able to cut a figure.

‘ numeric computation ‘ be ecbolic computer only then the person that make tomb figure, look from scientific point of view, border of calculative of numerical value of its reason nothing is more... than is exceeded not easily (Newton) kinetic system, have the law but abide, suit to be conveyed with machine computation. The main research direction of inchoate intelligence problem is the problem of rich play chess of chess card and so on, and the element limits that rich weichis is by demarcate inside a gather, its operation is the assorted change of all sorts of element only, the result also won‘t exceed border. Look so, train of thought of rich play chess and numeric computation have homogeneity. Problem of rich play chess considers to be in all previous classics after how many century, 1997 world chess king is opposite play chess IBM is deep blue system is pushed to acme, visible much boy comes computer expert is very happy with it all the time.

Is problem of rich play chess ‘ artificial intelligence ‘ main a home to return to? Computer researcher does not satisfy this kind of state, 70 time metaphase, the Fei Genbao Mu of Stanford university gave out in lecture will ‘ artificial intelligence ‘ turn the appeal to actual application: "Everybody is studying Bo Yiwen is inscribed... , ask everybody to cross real world, the problem in solving real world hard! The problem in solving real world hard!!

Solving medical treatment to diagnose a problem is a big department that artificial intelligence development comes out. 1972 De Domabl research and development ‘ acute bellyacke differentiates diagnostic system ‘ , shortliffe was finished 1976 ‘ contagious disease differentiates diagnostic system MYCIN ‘ etc. The research of likeness of bound of Chinese traditional Chinese medical science also is begun since 80 time like a raging fire, the system of expert of doctor of traditional Chinese medicine that about 140 experience are given priority to in succession research and development, those who use is the binary tree alls over all previous, backdate more recursive the algorithm of and so on and generation type regulation. A very serious problem is: In pattern of rich play chess although have Protean, but its castle ‘ soldier is soldier, horse it is a horse ‘ won‘t change; And the diagnostic element in diagnostic problem produces change possibly at any time. Get the expert system that train of thought of inchoate rich play chess affects so, in wanting to answer real world Protean it is very difficult.

2, arrive from expert system knowledge is decision-making

The medical treatment of real world diagnoses a problem to be different from rich play chess of course: It is nonexistent a strict Newton is kinetic the rule, without the border on dimensional meaning; The incident that it pays close attention to shows dynamic change process, change assumes random fluctuate position, the process cannot be gone against and cannot repeat, without the border on time meaning; Its element gather is not affirmatory, each element has his metabolic contrail, in-house existence restricts a mechanism in coordination, but cooperate with the mechanism of the mechanism is ambiguous; Solve the problem of diagnostic respect, be hope process change of course is final can convergence solves a dot to go up to, and it is the correlation that incline to diverge is a variety of likelihoods however actually, normally the doctor can intervene diagnostic result astringent arrives a certain a little bit forcibly.

Apparent, the problem arrived deep the academic category of diagnostic knowledge of human understanding mode and medicine. Want those who do away with train of thought of rich play chess to confuse bureau, turn artificial intelligence to practical, only way is to enter the intellectual field with extensive embroil. Fei Genbao Mu put forward timelily ‘ knowledge base ‘ , ‘ knowledge is conveyed ‘ and ‘ knowledge engineering ‘ wait for a series of concepts, from ‘ expert ‘ advanced state builds artificial intelligence conception. Fei Genbao Mu is described " one already is gifted the computer program of knowledge and ability, make this kind of action that reachs since program place reachs the expert‘s level thereby. Make this kind of action that reachs since program place reachs the expert‘s level thereby..

So ‘ expert ‘ what concept be? According to we normally understand: Master the knowledge of a course deep at least, have rich practice experience to this course. The expert‘s characteristic is ‘ experience ‘ , experience is a certain person is handling the specific event inside some domain repeatedly, because of a few detail that adroitness can bring about oversight or simplify, the action since these detail is to this person not character is axiomatic. We from turn over direction to unscramble ‘ expert ‘ : The expert‘s knowledge is finite, the expert‘s experience is individuation. Then we can think: ‘ expert ‘ the problem that can solve has probable sex and limitation only. If use the computer to handle ‘ expert ‘ problem, so, the expert‘s knowledge wants can very change conveniently to store condition, the expert‘s experience is inside expert individual limits, agree with the expert‘s knowledge, namely: Experience ＝ knowledge, can regard knowledge as processing next.

Fei Genbao Mu ‘ expert system ‘ concept and Newton are kinetic deterministic principle is not quite harmonious: If ‘ expert ‘ it is to point to an expert, criterion experience and the consistency that knowledge conveys do not have a problem, computer processing is more convenient also. But the knowledge of individual expert can solve the problem inside the domain to be restricted apparently, make expert system cannot seek solution to a lot of problems. The expert that this is train of thought of rich play chess fastens the main reason that all fails. If ‘ expert ‘ it is to point to many experts, want to solve the intellectual difference problem between the expert, this need undertakes to knowledge the standard is handled, form unified notional dictionary. About the expert ‘ people ‘ individual experience, one share rises to share knowledge, one share drops to abandan knowledge. Although can solve the consistency problem that systematic knowledge conveys so, but of expert experience broken be short of, brought about expert system to break away from original intention. Then ‘ rich experience of individuation ‘ decay is ‘ general character changes repeat knowledge ‘ , ‘ experience expert system ‘ decay is ‘ knowledge is decision-making system ‘ ; Earning wins the support of theory of abide card medicine more easily to conclusion, leave however go gradually with the deterministic requirement on kinetic meaning gradually far.

A lot of diagnosis of 80 time decision-making support system is developed by the train of thought of much expert system, among them two the famousest systems, one is still the ILIAD of his university, its knowledge base is to be below large hospital condition, sum up experience by the clinician, the knowledge that presses make it of Bayes model condition again is decision-making modular group; One is the Quick Medical Reference (QMR) of Pittsburgh university, its knowledge base is the arranges experience to handle medical literature according to system of expert of the medicine on large computer, exceed the decision-making model group that link means makes with what knowledge nods. Of two systems ‘ expert experience ‘ , rise for ‘ systematic knowledge ‘ , two systems should be belonged to " knowledge is decision-making system " .

(unfinished to be continued)