METHOD FOR DIAGNOSING DISEASES AND THEIR FORMS IN COMPUTER-ASSISTED MODE Russian patent published in 2002 - IPC

FIELD: medicine. SUBSTANCE: method involves building a system of heuristic characteristics describing various diseases, their classification on each disease to be diagnosed. Differential diagnosis of disease forms is found by applying cluster analysis methods in the space of heuristic characteristics. Building the system of heuristic characteristics involves selecting informative signs correlating to human organism states characteristic of particular disease within an optimum discretization interval and gradation scale to be determined for each quantitative sign with their following differentiation by disease forms. Classification on each disease to be diagnosed is carried out by means of integral comparison of disease signal to normal state threshold. Reliable disease recognition probability estimation P^* is determinable from a relation , where G<Mv >0 is the ratio of disease/normal state characteristic information capacities, G_i, is probability integral, N is the number of the characteristics, n_i is the current values of hardware signal/noise ratio and qualitative disease characteristic, respectively. EFFECT: higher accuracy of diagnosis in situations of lack of optimality and statistical insufficiency of medical characteristic database. 3 dwg, 1 tbl