Artificial Neural Networks for Obtaining New Medical Knowledge: Diagnostics and Prediction of Cardiovascular Disease Progression

摘要

Objectives: Development of a mathematical model and its implementation as a computer program for diagnosticsand prediction of progression of the most widespread cardiovascular diseases; the program is to model different variantsof disease progression for an observed patient, and to select individual recommendations for correction of his orher lifestyle, regimen and diet. Methods: Combination of technologies of neural networks and expert systems with aresulting synergistic effect. Results and Conclusion: Investigations of the developed mathematical model showed thatit is able to reveal new knowledge which is yet unknown to medical science. In the course of software experiments performedby means of a diagnostics-and-prediction system, we revealed the facts showing that modern medical practicepatterns of giving one and the same recommendations to all the cardiac patients without exception (including keepingto a hypocholesteric diet, giving up pernicious habits, limiting coffee and alcoholic drinks, losing weight and limitingintellectual and physical activity) are not always correct. Our investigations showed that some of these recommendationsare not just unhealthy, but harmful for a number of patients. The neuro-expert diagnostics-and-prediction systempresented in this paper allows doctors to reveal such non-typical patients and to develop individual recommendationsespecially for them.