2017 № 3 Analysis and forecasting of socially dangerous acts committed by psychiatric patients by dint of modern mathematical methods
The article considers the possibility of analyzing and forecasting socially dangerous acts committed by psychiatric patients by dint of modern mathematical methods. The review of modern methods of analysis and forecasting of time series is presented. An algorithm for forecasting socially dangerous acts committed by psychiatric patients according to the data of the regional forensic psychiatric expert service of the Tyumen region for the last 16 years has been developed with a total of 17967 examinations and an example of its program implementation is presented. Considered eight times of the series, the optimal model of which is selected from 188 in no less than 30 seconds. To assess the adequacy of the forecast, the model data obtained on the basis of the 15-year time series with the actual ones for 2016 were compared. All the mathematical models obtained confirmed the absence of significant differences in the Mann-Whitney U-test at significance level p = 0,05. Examples of interpretation of possible forecasts and examples of scientifically grounded management decisions are given.
2017 № 3 Portrait method as technology for development of decision making systems for diagnostics and differential diagnostics in clinical practice
Among the various technologies for the development of expert systems, the portrait method, originally pro- posed by E.V. Gubler. The authors have long used a portrait method for formalizing medical information when creating expert systems for diagnosis and differential diagnosis of diseases. When forming the knowledge base, the clinico-log- ical, expert and clinical-expert approach in the formation of a portrait of the disease is used. This article is enlightened on the description of the methodology for applying the portrait method in the development of expert systems. The features of forming the knowledge base and the algorithm of the solver work are considered
2017 № 2 Automation of differential diagnostics of craniocerebral trauma.
The authors are proposed a model of create expert systems to support decision making in emergency medical Neurology and Neurosurgery. The model assumes a symbiosis of clinical and expert approaches in the formation of the attribute space.
2016 № 5 Results of the expert systems comparison for diagnostics of acute pharyngeal diseases.
The description of the development of expert system (ES) of differential diagnostics of acute pharyngeal diseases based on different mathematical algorithms; «Portrait method», «Naiv Bayes classificator» (NBC), «Artificial Neuron Nets» (ANN) is presented in this article. Randonization of 476 of completed clinical cases with acute pharyngeal diseases, paratonsilitis, parapharyngitis, acute tonsillitis, acute pharyngitis is used. The technology of ES development is described. The clinical evaluation showed that ES based on «Portrait method» gives wrong diagnoses for whole list of diagnosed diseases by 8.40%. ES checking based on NBC in the clinical practice revealed differences in diagnoses in 47,6% of cases but developed ES «Program using for differential diagnosis of acute pharyngeal diseases «LOR-Neuro» showed high clinical efficiency in 96% of cases.
2015 № 1 Predicting the risk of postoperative complications in traumatic injuries of the pancreas.
The article is devoted to the problem of surgical treatment of traumatic injuries of the pancreas. Namely, identify significant risk factors for postoperative complications based on multivariate statistical analysis by logistic regression. The authors compiled prediction model specific postoperative complications with high rates of consent and clinical significance.