SYSTEM AND METHOD FOR DETERMINING SUBJECT CONDITIONS IN MOBILE HEALTH CLINICAL TRIALS

• US 20160140442A1
• Filed: 02/24/2015
• Published: 05/19/2016
• Est. Priority Date: 11/14/2014
• Status: Active Grant

• Current Assignee: Medidata Solutions Incorporated (Dassault Systemes SA)
• Original Assignee: Medidata Solutions Incorporated (Dassault Systemes SA)
• Inventors: Dennis, Kara, Lee, David, Savage, John
• Granted Patent: US 11,804,287 B2
• Time in Patent Office: Days
• US Class Current: 1/1
• CPC Class Codes: G06N 20/00 Machine learning
• CPC Class Codes: G06N 5/04 Inference or reasoning models
• CPC Class Codes: G16H 10/20 for electronic clinical trials or questionnaires
• CPC Class Codes: G16H 50/20 for computer-aided diagnosis, e.g. based on medical expert systems

Abstract

A method for calculating a subject'"'"'s state or condition comprises integrating data that are captured from multiple sources, storing the integrated data in a first database, calculating time intervals in which to collect an optimal amount of data to predict the subject state, developing a predictive model using a recorded diary or electronic data capture information, testing the model against a portion of the captured data, and applying the predictive model to new data from other sources. The predictive model may determine a subject state that may be a digital bio-marker for a disease condition. A system for predicting subject state is also disclosed.

Claims

1. A computer-implemented method comprising:

integrating data that are captured from multiple sources, wherein said data are associated with multiple variables that are evaluated to predict a state of a subject in a clinical trial;

storing the integrated data in a first database;

calculating time intervals in which to collect an optimal amount of said data to predict said subject state;

partitioning said integrated data by classifying a combination of ranges of values for the variables that optimally predicts the subject state;

generating a predictive model based on a classification algorithm to classify said state;

selecting an optimal data source to determine said subject state;

predicting a subject state by using said generated predictive model; and

storing the generated predictive model in a second database.

2. The method of claim 1, wherein the generating a predictive model based on a classification algorithm comprises using a machine-learning algorithm.

3. The method of claim 1, wherein the generating a predictive model based on a classification algorithm comprises using statistical analysis of the partitioned data.

4. The method of claim 1, wherein the predictive model calculates a subject state that is a digital bio-marker for a disease condition.

5. A system comprising:

a processor;

a model training module configured to receive data from a plurality of mobile health sources and to generate, using the processor, a predictive model of the state of a subject of a clinical trial based on said data and a data source that records said subject state; and

a scoring module configured to use the predictive model to calculate, using the processor, subject state for data without the data source that records said subject state.

6. The system of claim 5, wherein the model training module comprises:

a data integrator for integrating data that are captured from the mobile health sources;

a data cleaner for cleaning the integrated data and for storing the cleaned data in a database;

a model trainer for receiving a portion of the cleaned data and for classifying a combination of ranges of value for variables received from the mobile health sources;

a state predictor for developing a predictive model of subject state based on the training portion of the data; and

a model tester for testing a second portion of the cleaned data using the predictive model and comparing the predicted subject states against data from the data source that records said subject state.

7. The system of claim 5, wherein the scoring module comprises:

a data integrator for integrating new data that are input to the scoring module;

a data cleaner for cleaning the integrated new data; and

a scoring engine for receiving the predictive model and the cleaned data and calculating a subject state.

8. The system of claim 5, wherein the plurality of mobile health sources come from a single mobile health device.