| Under the background of population aging,the number of patients with Alzheimer’s disease has increased rapidly.Because of its care difficulties and huge human and mental investment,in the case of uneven distribution of medical resources,it has greatly affected the lives of patients and caregivers,and brought a heavy burden to their families and the whole society.As a major way to improve the shortage of medical resources,telemedicine service is not limited by time,space and location,and has broad development prospects.So it is of great practical significance to study and design a telemedicine consultation system for dementia patients.With the rapid development of the "Internet + medical" model,many online consultation platforms can be found in the market,including those based on web pages and those based on mobile terminals.On the mobile terminal,the mainstream remote consultation platforms include "Ding Xiang Doctor","Jing Dong Health","Good Doctor",etc.,when collecting the symptoms,patients are mostly asked to input them,the previous selection can not be used to get more recommendation for it.In this thesis,the inquiry system is designed for dementia research,the symptoms sent to patients should be more professional,so as to facilitate the diagnosis of doctors and improve the quality of inquiry.To improve the efficiency of diagnosis and experience of interactive,more similar symptoms are recommended to users on the basis of user selection,so as to enrich and complete the disease descriptions.Therefore,based on dementia,in-depth research on the intelligent symptom recommendation algorithm is carried on in this thesis.The specific work contents are as follows:(1)Analyze the shortcomings and deficiencies of traditional collaborative filtering recommendation algorithms and a deep symptom recommendation algorithm that combine attention and knowledge graphs is researched and implemented.According to the research field and needs of the subject,web crawler was used to crawl the data of neurology and related fields,and a domain knowledge map based on dementia was constructed.The existing deep recommendation algorithms are compared and their advantages and disadvantages are analyzed through experiments.By designed and improved and the feature extraction module is designed and improved on the basis of the MKR model then AFM-MKR model is proposed.The symptom recommendation data set is constructed by preprocessing data from hospital.By use of deep neural network,taken symptoms as the target of training and output,the potential associations between patients’ symptoms with the user’s name,age,gender,ethnicity,home address,past history,allergy history,family history,biography and seasonal factors were mined.The symptoms that patients might choose are sent to the user according to the probability prediction model.Experiments show that the model in this thesis achieves 62.49%,62.73%,and 62.61% in accuracy,recall,and F1-score,respectively.(2)Research and implementation of dynamic symptom recommendation algorithm based on knowledge graph representation learning algorithm.According to the characteristics of dementia,a symptom recommendation algorithm based on knowledge graph representation learning is determined.Experiments show that the symptom recommendation algorithm model based on knowledge graph representation learning can well solve the recommendation of more similar symptoms based on previous choices,improve system interaction and diversity of symptom recommendations.(3)Design and implementation of telemedicine consultation system for dementia patients.The actual needs of the system are analyzed,the above-established symptom recommendation algorithm is used to design and implement a patient-side consultation system for the telemedicine and comprehensive care integrated prevention and control system. |
PDF Full Text Request