Research On Collaborative Drugs Mining Method For Reducing Hospitalization Time Of Patients With Traumatic Brain Injury

With the development of contemporary information technology,various artificial intelligence methods are applied in research related to the fields of medicine and biology.In the face of the development of big data technology and storage technology,a large amount of clinical data has been generated.This massive data mining and analysis is a problem worthy of research.Moreover,the clinical treatment of many diseases requires the use of a large number of drugs at the same time,but the potential synergy and adverse events associated with multi-drugs are still unclear.For example,it is not clear which drugs have a decisive effect on the treatment of traumatic brain injury,which drugs are of higher importance,and which drugs are used in conjunction with side effects.Currently,there is a lack of research to systematically study this challenge.In this paper,starting from diseases without specific drugs,we propose a method of data mining for diseases without specific drugs to discover single or multiple drugs that can significantly reduce the length of hospitalization for patients.Taking traumatic brain injury as the main research disease,we sorted out the research situation of traumatic brain injury disease at home and abroad,and used the data of traumatic brain injury patients in a hospital to conduct our research.In the study of a single drug that can reduce the length of hospital stay in patients with traumatic brain injury,starting from the generally observed phenomenon,the longer the hospital stay,the higher the frequency of certain drugs.We use the exhaustive method,that is,sort by the overall frequency of use of each drug in the data set,and use biostatistics survival analysis to classify the drugs that are higher than the average use frequency of a certain drug and lower than the average use frequency,Using the length of stay in the two types of patients and whether they are discharged as basic data for survival analysis.When the length of hospitalization of patients who use a drug with high frequency is significantly lower than that of patients who use a drug with low frequency,we believe that this drug can reduce the length of hospitalization for traumatic brain injury.When mining two kinds of cooperative drugs that can reduce the length of hospitalization of patients with traumatic brain injury,we use neural network to embed the corresponding word vector,and after obtaining the word vector,form a sentence vector according to the hospitalization drug prescription of each patient.Using the dimensionality reduction method and clustering algorithm in machine learning,the difference between survival analysis and survival distribution is used to find the difference in the frequency of paired medications in the two classes with the largest difference.Compare the size of the difference in the frequency of the two types of medications,and rank them.The use of the high-frequency combined use of the two drugs in the hospitalization time collection and the low frequency combined use of the two drugs in the hospitalization time collection determines that it can reduce the traumatic brain injury patients Two drugs used in conjunction with the length of the hospital stay.The final results show that our method can reduce the length of hospital stay of patients with traumatic brain injury by mining in the first 3000 cooperatively used drugs,compared with the method of sorting and searching according to the frequency of cooperatively used drugs on the entire data set.At the same time,these excavated drugs are also worthy of further research by clinical experts.This method is based on the purpose of exploring two drugs that are used together.Further research and improvement are needed to further study whether the coordinated use of three or more drugs can reduce the length of stay in patients with traumatic brain injury.