Medical data processing and system based on migration learning

Abstract

The invention discloses a medical data processing and system based on migration learning. The medical data processing comprises the following steps: acquiring text data outside the medical field, andtraining to obtain a text classification model; acquiring a case set in the medical field, wherein the case set comprises symptoms and labels, and the labels are symptoms corresponding to the diseasesymptoms; extracting the characteristic vectors of the symptoms by using the text classification model as the symptom vectors, and converting the labels into label vectors according to the disease symptom types corresponding to the symptoms; constructing a multi-label training sample set by integrating the symptom vectors and the corresponding label vectors, and training to obtain a multi-label classification model according to the multi-label training sample set; inputting the medical samples to be analyzed into the multi-label classification model, determining the probability values of the medical samples belonging to each type of labels, and obtaining an analysis label set according to the probability values to serve as the analysis result of the medical samples. Therefore, the defect of manual selection of features is avoided through migration learning, and the medical disease prediction accuracy based on outpatient cases is improved.

Description

technical field [0001] The invention relates to the field of medical disease analysis and machine learning, in particular to a medical data processing and system based on migration learning. Background technique [0002] With the continuous development and wide application of artificial intelligence and machine learning, a clinical auxiliary decision support system is proposed to predict and analyze the condition based on patient information and rely on machine learning analysis models, which can help doctors to be more efficient in the process of clinical analysis and decision-making , More quickly apply complex medical knowledge to deal with various medical problems. [0003] Medical disease prediction model is one of the core challenges of intelligent auxiliary analysis system, which can be divided into rule-based expert model, statistical analysis model based on statistical knowledge and prediction model based on machine learning. [0004] The paper (Shortliffe E.H.Comp...

AI Technical Summary

Problems solved by technology

However, some medical diseases are characterized by a small number of samples, that is, the number of cases of the disease is relatively small, or the collection is so difficult that it is difficult to train an ideal machine learning model

[0007] Some medical diseases have problems of collection difficulty and low incidence, which makes it difficult to have a large number of samples for deep neural network training, so it is difficult to train a model with strong analysis performance in this field

Images