Applying methods of twin comparing quantitative and binary samples in biomedical information systems for decision making

Solving research problems within the framework of creating a single digital circuit in healthcare requires a research conducted on the basis of depersonalized medical data stored in the information systems of medical institutions. There are described the methods of mathematical statistics aimed at comparing the average values of two types of samples: quantitative and binary in order to determine the relationship between blood test indicators and working conditions. Comparison of methods and results of comparison of quantitative and binary samples is made. The expediency of processing small structured samples taken out from the medical information system is substantiated. The study was conducted by using medical data stored in a transactional medical information system. During the preparation process, the data were depersonalized, cleaned from the inevitable noise and defects. Binarization of the values of the indicators was performed by comparing them with the known boundaries of the interval of the medical norm. A method was developed to bring the samples to uniformity simultaneously on the gender and age signs of the patients. There have been revealed the parameters of laboratory tests, which have a statistically significant relationship with working conditions identified for 4 groups under study. These groups were corresponding to the following work conditions complexes: influence of electromagnetic emanation, noise and vibrations, working conditions in regional office services. The proposed methods and received results will increase the accuracy of the performed risk assessments of occupational morbidity and become the base for studying the mechanism of the work conditions influencing the health. They will contribute to improvement of the analysis of the data collected in the medical information systems and management decision-making in healthcare.