Auxiliary Diagnosis And Drug Decision-Making Of Chronic Diseases In The Elderly Based On Deep Learning

In developing countries,the problem of population aging is becoming increasingly prominent,and chronic diseases,such as diabetes,hypertension,and pulmonary nodules,constitute the most important threat to the aging society.Due to limited medical resources and large population,hospitals cannot provide timely treatment for the elderly quickly,and the diagnosis and treatment of chronic diseases in the elderly is extremely challenging.In addition,in the process of chronic disease treatment,long-term and high treatment costs have become a heavy economic burden for patients.The research on auxiliary diagnosis of chronic diseases is essentially the use of deep learning technology to mine various rules and connections of chronic disease diagnosis,and assist doctors in diagnosis through effective data decision-making and wireless network communication.Aiming at the problem that classification models cannot correlate wearable device monitoring data and process multi-source data,this thesis proposes a combined sparse autoencoder model for chronic disease auxiliary diagnosis.The model uses sparse autoencoders to process patient detection data and real-time data such as blood pressure,heart rate,and blood sugar monitored by wearable devices collaboratively to achieve multi-source data association and fusion.After backpropagation algorithm training and softmax classifier output disease diagnostic probability,the best disease diagnostic sequence is obtained to realize warning of chronic diseases.Four chronic disease datasets,including heart disease,diabetes,chronic kidney disease,and hypertension,are used as experimental data.The combined sparse autoencoder and four classification algorithms,including Naive Bayes,k-nearest neighbor,linear discriminant analysis,and artificial neural network,are used as experimental comparisons.The experimental results show that the proposed method performs better in accuracy and sensitivity.Aiming at the medical cost in the treatment of elderly patients with pulmonary nodules,considering the differences in the economic ability of different patient groups and the individual requirements of different patient groups,this thesis presents a neural network model for drug decision-making in pulmonary nodules.Through the artificial neural network model,the effectiveness of drugs for the treatment of pulmonary nodules is evaluated through the changes in the area,density and number of pulmonary nodules,and the drug cost is used as auxiliary decisionmaking information to conduct drug effectiveness-cost analysis.From the perspective of probabilistic decision-making,the ratio of the incremental effectiveness of the drug to the incremental cost is analyzed,and the optimal therapeutic drug sequence is output to complete the drug decision-making judgment.Patients with pulmonary nodules taking six different drugs are selected as experimental data.The experimental results show that the method proposed in this thesis can rely on the clinical data provided by the hospital to decide the drug treatment plan suitable for the patient’s condition,saving medical resources and costs.