An application of stochastic estimation to diagnostic of hereditary metabolism diseases (Q1282570)

Some approaches to the design of algorithms of computer-aided diagnostics are presented. Computer-aided diagnostics is used for diagnostics of hereditary metabolism diseases and the algorithms are based on methods of stochastic estimation. Let a certain event \(Y\) mean that ``a hereditary metabolism disease takes place''. \(Y\) is a union of independent events \(Y_1,\dots, Y_m\), where \(Y_i\) means ``disease \(i\) takes place''. The a posteriori probabilities of disease \(P(Y_i| S^*)\), \(i= 1,\dots, m\), are estimated, where \(S^*= (S^*_1,\dots, S^*_n)\) is an observed set of clinical symptoms \(S_1,\dots, S_n\). For this purpose point estimators of a priori probabilities \(P(Y_i)\), \(P(S_j| Y_i)\), \(i=1,\dots, m\), \(j= 1,\dots, n\) are used. Two practical methods of diagnostics are considered. The first one is within the scope of the stochastic estimation scheme, and the second one is based on the probabilistic guaranteed approach. Results on the efficiency estimation of the suggested algorithms are given.