Search academic texts

Information × Registration Number 0221U105885, 0121U111557 , R & D reports Title Decision support system for modeling the spread of viral infections popup.stage_title Head Shakhovska Nataliia B., д.т.н. Registration Date 07-09-2021 Organization Lviv Polytechnic National University popup.description2 A model of conditional multidimensional space for personalized medical data is designed. This allows to predict the value of target variables for certain space measurements with higher accuracy. A new ensemble for predicting inpatient duration has been developed treatment based on a hierarchical predictor with 5-fold validation results based on a two-stage data analysis procedure: clustering and forecasting for each separate cluster. It gives in 1.47 times better prediction accuracy than the best weak predictor (perceptron with 12 units in one hidden layer). The complexity of the parallel ensemble and the ensemble without paralelization is determined and calculated the dependence of the reduction of time complexity in depending on the number of cores. This allows to apply the developed models for processing large data sets. A new inputs doubling method based on use classical iterative RBF neural network has been developed, which improves the accuracy of solving the prediction task in case of short data sets. A set of data with information about personal parameters of patients is formed, which allowed to test the developed methods and algorithms.   Product Description popup.authors Izonin Ivan V Vyklyuk Yaroslav I Kryvenchuk Yurii P Melnykova Nataliia I popup.nrat_date 2021-09-07 Close