Study of the effectiveness of individual computer-assisted prediction of the results of rehabilitation treatment of patients with schizophrenia
Burkovskiĭ, G.V.; Vuks, A.I.; Iovlev, B.V.; Korabel'nikov, K.V.; Shchelkova, O.I.
Zhurnal Nevropatologii i Psikhiatrii Imeni S.S. Korsakova 90(7): 94-99
1990
ISSN/ISBN: 0044-4588 PMID: 2175102 Document Number: 356149
The rehabilitation automated information system was used to analyze the data of 768 formalized case reports of schizophrenia patients with a purpose of elaboration and control of the efficacy of the computing rules of individual prediction of the remission pattern. Two algorithms of image recognition were employed: a nonuniform statistical procedure and Bayes++'s method. A study was made of the relationship between the prediction accuracy and the number of traits (from 10 to 86), ranged on the basis of prognostic information content according to the Kulbak's measure. Using 86 traits with the aid of Bayes' method, 76% of correct prognoses, 20% of erroneous and 4% of indefinite ones were obtained. The computing prediction tables were designed, intended for practicing physicians, requiring no use of computers.