Identifying and indexing discriminative features for disease progression in observational data

Abstract

A system (or method) for generation and employment of disease progression model(s) that facilitates identifying and indexing discriminative features for disease progression in observational data. The disease progression prediction system comprises a processor that executes computer executable components stored in memory. A receiving component receives and learns observational patient data. A model generation component builds a preliminary disease progression model. An identification component identifies discriminative clinical features for different disease stages. A ranking component ranks discriminative powers of clinical features for respective pairs of disease stages; wherein the model generation component employs the ranked features to generate a final disease progression model.

What is claimed is: 
     
         1 . A system, comprising:
 a memory that stores computer executable components;   a processor, operably coupled to the memory, that executes computer executable components stored in the memory, wherein the computer executable components comprise:
 a model generation component that builds a preliminary disease progression model based on observational patient data; and 
 a ranking component that ranks discriminative powers of clinical features for respective pairs of disease stages, wherein the model generation component employs the ranked discriminative powers of clinical features to generate a final disease progression model, wherein generation of the final disease progression model comprises by the model generation component:
 determination of component features for data-driven disease stage segmentation; 
 assignment of respective observations to one of the disease stages to generate disease progression modeling resulting in disease stage assignment; and 
 for respective ones of the clinical features in the observational patient data, employing assigned disease stages as benchmarks to obtain effective sizes of the clinical features for the respective pairs of disease stages. 
 
   
     
     
         2 . The system of  claim 1 , wherein the computer executable components further comprise a filtering component that combines existing medical knowledge of a target disease as well as availability of clinical features in the observational patient data to perform an initial feature filtering. 
     
     
         3 . The system of  claim 2 , wherein the filter component filters features most irrelevant to disease progression to generate a reduced dataset. 
     
     
         4 . The system of  claim 3 , wherein the identification component performs composite feature engineering to identify underlying disease progression directions from the reduced dataset. 
     
     
         5 . The system of  claim 1 , wherein the generation of the final disease progression model further comprises assignment of respective observations to one of the disease stages employing a semi-Markov jump process to generate disease progression modeling resulting in disease stage assignment. 
     
     
         6 . The system of  claim 5 , wherein the clinical features are time stamps for respective patients showing disease progression thereby creating longitudinal data that is specific for respective patients. 
     
     
         7 . The system of  claim 6 , wherein the computer executable components further comprise a pooling component that, for respective pairs of disease stages, pools clinical features and the ranking component ranks the clinical features by respective effective size. 
     
     
         8 . The system of  claim 1 , wherein the final disease progression model contains clinical features effective for discriminating respective pairs of disease stages. 
     
     
         9 . The system of  claim 1 , wherein the final disease progression model employs a utility-based analysis to factor benefit of making a correct prediction against cost of making an incorrect prediction. 
     
     
         10 . A computer-implemented method, comprising:
 building, by a system operatively coupled to a processor, a preliminary disease progression model based on observational patient data;   ranking, by the system, discriminative powers of clinical features for respective pairs of disease stages; and   employing, by the system, the ranked discriminative powers of clinical features to generate a final disease progression model, wherein generation of the final disease progression model comprises:
 determination of component features for data-driven disease stage segmentation; 
 assignment of respective observations to one of the disease stages to generate disease progression modeling resulting in disease stage assignment; and 
 for respective ones of the clinical features in the observational patient data, employing assigned disease stages as benchmarks to obtain effective sizes of the clinical features for the respective pairs of disease stages. 
   
     
     
         11 . The computer-implemented method of  claim 10 , further comprising combining existing medical knowledge of a target disease as well as availability of clinical features in the observational patient data to perform an initial feature filtering. 
     
     
         12 . The computer-implemented method of  claim 11 , further comprising filtering features most irrelevant to disease progression to generate a reduced dataset. 
     
     
         13 . The computer-implemented method of  claim 12 , further comprising performing a composite feature engineering step to identify underlying disease progression directions from the reduced dataset. 
     
     
         14 . The computer-implemented method of  claim 13 , wherein the final disease progression model is built based as a function of composite features for data-driven disease stage segmentation, and assignment of respective observations to a disease stage. 
     
     
         15 . The computer-implemented method of  claim 14 , further comprising for respective pairs of disease stages, clinical features are pooled and the clinical features are ranked by effective size. 
     
     
         16 . The computer-implemented method of  claim 10 , wherein the final disease progression model contains clinical features effective for discriminating respective pairs of disease stages. 
     
     
         17 . The computer-implemented method of  claim 16 , further comprising training the final disease progression model using recursive machine learning. 
     
     
         18 . The computer-implemented method of  claim 17 , employing, by the system, utilizing the final disease progression model, a utility-based analysis to factor benefit of making a correct prediction against cost of making an incorrect prediction. 
     
     
         19 . A computer program product for monitoring disease progression, the computer program product comprising a computer readable storage medium having program instructions embodied therewith, the program instructions executable by a processor to cause the processor to:
 build a preliminary disease progression model based on observational patient data; and   rank discriminative powers of clinical features for respective pairs of disease stages, wherein the model generation component employs the ranked discriminative powers of clinical features to generate a final disease progression model, wherein generation of the final disease progression model comprises by the model generation component:
 determination of component features for data-driven disease stage segmentation; 
 assignment of respective observations to one of the disease stages to generate disease progression modeling resulting in disease stage assignment; and 
 for respective ones of the clinical features in the observational patient data, employing assigned disease stages as benchmarks to obtain effective sizes of the clinical features for the respective pairs of disease stages. 
   
     
     
         20 . The computer program product of  claim 19 , wherein the program instructions are further executable by the processor to cause the processor to:
 combine existing medical knowledge of a target disease as well as availability of clinical features in the observational patient data to perform an initial feature filtering.