Medication analysis method based on full memory event sequence mining model

Abstract

The invention belongs to the technical field of data mining, medical information and big data, and particularly to a medicine analysis method based on full memory event sequence mining model. The method converts original medication datum into a medication event occurrence sequence with category type, converts each medication event node of the medication event occurrence sequence with category type into a multidimensional vector representation of euclid space, and then weighted summation is conducted for all the historical events according to the previous node of the medication event to be predicted to form a character representation of the memory of a predict event and served as a standard input of a classifier for predicting the next event vector, the multidimensional vector representation of the event to be predicted is served as an output, a predictive model is trained, the multidimensional vector of euclid space is mapped back to the original space with category type, all the medication event sequences are trained, finally the models trained are input to predict future events for the new medication event sequence. The method can use data as many as possible for decision-making, thereby reducing decision-making mistakes.

Description

Technical field [0001] The invention belongs to the technical fields of data mining, medical information and big data, and specifically relates to a medication analysis method based on a full memory event sequence mining model. Background technique [0002] An effective medication regimen is very important for patients to obtain the best treatment. Using the accumulated large-scale historical data of patients' clinical medications for analysis and modeling will help doctors provide decision support for patients' next medications. Data mining methods have been used for medication analysis. Traditional medication data mining methods use frequent pattern mining to discover frequently occurring medication combinations and infer the association between medications. However, on the one hand, frequent pattern mining that does not consider the order of medication ignores the influence of the characteristics of disease evolution on medication; on the other hand, although frequent sequenc...

AI Technical Summary

Problems solved by technology

However, on the one hand, frequent pattern mining that does not consider the order of medication ignores the impact of disease evolution characteristics on medication; on the other hand, although frequent sequence pattern mining methods that consider "order" and time characteristics have been used, however, a large number of Redundant result schemas and association rules limit the practical application of the method

Images