Method and apparatus for analyzing medical treatment data based on deep learning

Abstract

Disclosed are a method and an apparatus for analyzing medical treatment data based on deep learning. The core content of the method is the establishment of a model in a computer using a deep convolution neuron algorithm in deep learning. The model assists the doctors to make correct judgments and effective decisions for the large amount of medical treatment data using mass medical treatment data selection and optimization model parameters by “training” the model to automatically learn a pathology analysis process of the doctors or the medical researchers and then helping them in processing the large amount of medical treatment data. The present invention can greatly reduce the work stress for the doctors or the medical researchers and improve the work efficiency thereof, and can free the doctors or the medical researchers from heavy analysis work on the medical treatment data or medical data.

What is claimed is: 
     
         1 . A method for analyzing medical treatment data based on deep learning, comprising the following steps:
 obtaining to-be-analyzed medical treatment data;   inputting the to-be-analyzed medical treatment data into a deep learning model to conduct a medical pathological analysis matched therewith, wherein the deep learning model is established based on data features of medical treatment training data, and the medical treatment training data comprise medical treatment raw data and medical treatment diagnosis data matched with the medical treatment raw data; and   outputting a medical pathological analysis result matched with the to-be-analyzed medical treatment data;   wherein the deep learning model comprises an input layer comprising a plurality of nodes having a plurality of extracted features of the medical treatment raw data comprising related records on patient diagnosis, detection and treatment by clinical doctors and medical technicians, and an output layer comprising a plurality of nodes with a plurality of extracted features of the medical treatment diagnosis data comprising related records on initial diagnosis, discharge result and disease treatment effect by clinical doctors and medical technicians, and textual visiting records and follow-up data by doctors; and   the data features of medical treatment training data comprise a spatial and temporal variation of a lesion scan during a process from initial diagnosis to disease treatment and then to hospital discharge.   
     
     
         2 . The method according to  claim 1 , wherein the deep learning model is trained to perform an intelligence diagnosis of pulmonary diseases. 
     
     
         3 . The method according to  claim 1 , further comprising:
 transforming or formatting the medical treatment training data into computer-understandable structured data matrix by segmentation, correlation or text data mining methods;   inputting the medical treatment training data formed as structured data matrix into the deep learning model;   optimizing the deep learning model, wherein the optimization method comprising the following steps:   a. constructing a primary deep learning frame to establish a data model comprising the input layer, at least a hidden layer and the output layer, according to the data features of the medical treatment training data, wherein each hidden layer comprises a plurality of nodes mapping with an output from a previous layer thereof;   b. constructing the data model for each of the nodes with a mathematical formula, wherein related parameters for the mathematical formula are preset manually or at random automatically, inputs to the nodes of the input layer are the data features of the medical treatment raw data, and inputs to respective hidden layer and the output layer are outputs generated at the previous layer thereof respectively, wherein the outputs at each node of each layer is obtained according to the mathematical formula thereof; and   c. initializing a parameter Ai and comparing the output generated at each node of the output layer with the medical treatment diagnosis data stored at the respective node, and modifying the parameter Ai at the node thereby in an orderly-cycling manner to ultimately obtain the parameter Ai at the respective node which enables the output generated at the respective node of the output layer at its partially maximum of the corresponding node similar to the data features of the medical treatment diagnosis data.   
     
     
         4 . The method according to  claim 3 , wherein a simulation data is constructed by deforming, distorting, and noise-superimposing the medical treatment training data before inputting the medical treatment training data into the corresponding deep learning model. 
     
     
         5 . The method according to  claim 1 , wherein structured data of the to-be-analyzed medical treatment data and medical analysis data matched therewith are fed back to the deep learning model as new training data. 
     
     
         6 . The method according to  claim 1 , wherein the medical pathological analysis result comprises an analysis of disease identification, treatment advice and proposed treatment project. 
     
     
         7 . The method according to  claim 1 , wherein the medical pathological analysis result comprises a type of lesion, the treatment effect and the specific location of the lesion. 
     
     
         8 . The method according to  claim 1 , wherein the deep learning model is trained to learn to predict the development of medical phenomena. 
     
     
         9 . The method according to  claim 3 , wherein the method to optimize the parameter Ai is an unsupervised learning method or a supervised learning method. 
     
     
         10 . An apparatus for analyzing medical treatment data based on deep learning, comprising: a processor; and a memory storing computer executable instructions, which when executed by the processor cause the processor to perform operations comprising:
 obtaining to-be-analyzed medical treatment data;   inputting the to-be-analyzed medical treatment data into a deep learning model to conduct a medical pathological analysis matched therewith, wherein the deep learning model is established based on data features of medical treatment training data, and the medical treatment training data comprise medical treatment raw data and medical treatment diagnosis data matched with the medical treatment raw data; and   outputting a medical pathological analysis result matched with the to-be-analyzed medical treatment data;   wherein the deep learning model comprises an input layer comprising a plurality of nodes having a plurality of extracted features of the medical treatment raw data comprising related records on patient diagnosis, detection and treatment by clinical doctors and medical technicians, and an output layer comprising a plurality of nodes with a plurality of extracted features of the medical treatment diagnosis data comprising related records on initial diagnosis, discharge result and disease treatment effect by clinical doctors and medical technicians, and textual visiting records and follow-up data by doctors; and   the data features of medical treatment training data comprise a spatial and temporal variation of a lesion scan during a process from initial diagnosis to disease treatment and then to hospital discharge.   
     
     
         11 . The apparatus according to  claim 10 , wherein the deep learning model is trained to perform an intelligence diagnosis of pulmonary diseases. 
     
     
         12 . The apparatus according to  claim 10 , wherein the operations further comprises:
 transforming or formatting the medical treatment training data into computer-understandable structured data matrix by segmentation, correlation or text data mining methods;   inputting the medical treatment training data formed as structured data matrix into the deep learning model;   optimizing the deep learning model, wherein the optimization method comprising the following steps:   a. constructing a primary deep learning frame to establish a data model comprising the input layer, at least a hidden layer and the output layer, according to the data features of the medical treatment training data, wherein each hidden layer comprises a plurality of nodes mapping with an output from a previous layer thereof;   b. constructing the data model for each of the nodes with a mathematical formula, wherein related parameters for the mathematical formula are preset manually or at random automatically, inputs to the nodes of the input layer are the data features of the medical treatment raw data, and inputs to respective hidden layer and the output layer are outputs generated at the previous layer thereof respectively, wherein the outputs at each node of each layer is obtained according to the mathematical formula thereof; and   c. initializing a parameter Ai and comparing the output generated at each node of the output layer with the medical treatment diagnosis data stored at the respective node, and modifying the parameter Ai at the node thereby in an orderly-cycling manner to ultimately obtain the parameter Ai at the respective node which enables the output generated at the respective node of the output layer at its partially maximum of the corresponding node similar to the data features of the medical treatment diagnosis data.   
     
     
         13 . The apparatus according to  claim 12 , wherein a simulation data is constructed by deforming, distorting, and noise-superimposing the medical treatment training data before inputting the medical treatment training data into the corresponding deep learning model. 
     
     
         14 . The apparatus according to  claim 10 , wherein structured data of the to-be-analyzed medical treatment data and medical analysis data matched therewith are fed back to the deep learning model as new training data. 
     
     
         15 . The apparatus according to  claim 10 , wherein the medical pathological analysis result comprises an analysis of disease identification, treatment advice and proposed treatment project. 
     
     
         16 . The apparatus according to  claim 10 , wherein the medical pathological analysis result comprises a type of lesion, the treatment effect and the specific location of the lesion. 
     
     
         17 . The apparatus according to  claim 10 , wherein the deep learning model is trained to learn to predict the development of medical phenomena. 
     
     
         18 . The apparatus according to  claim 12 , wherein the method to optimize the parameter Ai is an unsupervised learning method or a supervised learning method. 
     
     
         19 . A non-transitory computer readable storage medium, wherein the storage medium stores computer programs, and the computer programs are configured to perform a method for analyzing medical treatment data based on deep learning, comprising:
 obtaining to-be-analyzed medical treatment data;   inputting the to-be-analyzed medical treatment data into a deep learning model to conduct a medical pathological analysis matched therewith, wherein the deep learning model is established based on data features of medical treatment training data, and the medical treatment training data comprise medical treatment raw data and medical treatment diagnosis data matched with the medical treatment raw data; and   outputting a medical pathological analysis result matched with the to-be-analyzed medical treatment data;   wherein the deep learning model comprises an input layer comprising a plurality of nodes having a plurality of extracted features of the medical treatment raw data comprising related records on patient diagnosis, detection and treatment by clinical doctors and medical technicians, and an output layer comprising a plurality of nodes with a plurality of extracted features of the medical treatment diagnosis data comprising related records on initial diagnosis, discharge result and disease treatment effect by clinical doctors and medical technicians, and textual visiting records and follow-up data by doctors; and   the data features of medical treatment training data comprise a spatial and temporal variation of a lesion scan during a process from initial diagnosis to disease treatment and then to hospital discharge.   
     
     
         20 . The non-transitory storage medium according to  claim 19 , wherein the deep learning model is trained to perform an intelligence diagnosis of pulmonary diseases.