Semi-supervised Learning-Based Decision-making Method Of Medical Order Aided

With the rapid development of Internet information storage and data processing technology,the hospital’s informatization construction has received effective technical support.And the medical records of patients have been changed from the paper records to the digital way of constructing the electronic medical record system.The electronic medical record data serves as an important carrier of information in the patient’s visit,and provides data support for intelligent auxiliary medical decision-making research.However,due to the variety of electronic medical record data,such as forms,numbers and free texts,unstructured text data contains a large number of specialized descriptions,etc.,and it also takes time and labor to preprocess electronic medical record data and mark data.And other issues.In order to more effectively tap the potential value of electronic medical records,and using the experience of existing clinical diagnosis and treatment,many scholars combined with machine learning technology to study medical decision-making methods.In view of the characteristics of the above electronic medical record data and the problems in the research process,combined with the application of the machine learning method in the less practical problem of labeling samples,this paper intends to adopt the semi-supervised learning method,based on the combination of a small amount of labeled data,from the patient similar group.Starting from two aspects of the choice of doctor’s advice,the doctor’s decision-making method is studied.It mainly includes the following three parts:(1)Data preprocessing research.This paper has normalized and standardized mapping for multi-type data of electronic medical records.At the same time,considering the large proportion of free texts and the professional characteristics of the content,combined with the construction of domain dictionary auxiliary word segmentation in text segmentation,the effectiveness of the method for medical terminology recognition is verified by comparative experiments.(2)Study of patient similar group classification.This paper comprehensively considers the patient basis and diagnosis information to construct the patient attribute system.At the same time,based on the multi-indicator consideration,the patient’s similarity degree estimation is used to construct the patient paired constraint set as the supervision information to guide the patient cluster learning model and optimize the similarity group classification effect of patients.(3)Research on doctor-assisted decision-making methods.This paper proposes that in the process of selecting the medical order list,the text clustering method is first used to classify the doctors according to the type of diagnosis and treatment plan,and comprehensively consider the similarity of patients and the influence of medical order on the choice of medical order.According to the similarity of patients and the importance of different treatment plans Sexuality,the overall correlation of each doctor’s order is calculated by a more fine-grained analysis of the two dimensions of the patient and the treatment plan.In this paper,through multiple sets of contrast experiments,it is proved that the similarity group classification based on semi-supervised learning is better than the simple unsupervised clustering algorithm,which comprehensively considers the multi-dimensional attribute information of electronic medical records and patient paired constraint information,and assists patients in the treatment plan category.The relationship between doctors and nurses can improve the effectiveness of the doctor-assisted decision-making model.The semi-supervised learning-based clustering method proposed in this study can be better applied to the research of medical data.Based on this,the doctor-assisted decision-making thought based on patient similarity,doctor’s advice and patient correlation analysis can help.Better use of existing data resources to assist clinical staff in the diagnosis and treatment,and at the same time adopt appropriate methods of diagnosis and treatment to improve the efficiency and accuracy of the prescription.