Use of web-based symptom checker data to predict incidence of a disease or disorder

• US 10,430,904 B2
• Filed: 02/14/2014
• Issued: 10/01/2019
• Est. Priority Date: 02/15/2013
• Status: Active Grant

• Current Assignee: Battelle Memorial Institute
• Original Assignee: Battelle Memorial Institute
• Inventors: McMillan, Nancy, Feng, Jingyu, Stamps, Kathryn, Burr, Robert E.
• Primary Examiner(s): Ng, Jonathan
• Application Number:
  US14/767,820
• Publication Number:
  US 20150371006A1
• Time in Patent Office: 2,055 Days
• CPC Class Codes: G06Q 10/04 Forecasting or optimisation specially adapted for administrative or management purposes, e.g. linear programming or "cutting stock problem" market predictions or forecasting for commercial activities G06Q30/0202
• CPC Class Codes: G16H 15/00 ICT specially adapted for medical reports, e.g. generation or transmission thereof
• CPC Class Codes: G16H 40/67 for remote operation
• CPC Class Codes: G16H 50/20 for computer-aided diagnosis, e.g. based on medical expert systems
• CPC Class Codes: G16H 50/80 for detecting, monitoring or modelling epidemics or pandemics, e.g. flu
• CPC Class Codes: Y02A 90/10 Information and communication technologies [ICT] supporting adaptation to climate change, e.g. for weather forecasting or climate simulation

Abstract

Symptoms and methods for predicting the incidence of a disease or disorder are disclosed. A system for predicting the incidence of a disease or disorder includes a web-based symptom checker for producing a structured dataset, a data analysis component for producing a multivariate dataset from the structured dataset, and a feature construction component for producing a linear combination of orthogonal symbols representative of a disease or disorder. A method for predicting the incidence of a disease or disorder includes producing a multivariate dataset representing patient symptom counts, performing feature construction analysis on the multivariate dataset, creating a time series model using weekly illness incidence data, and applying the time series model to new illness incidence data to predict the incidence of a disease or disorder in the future.

Claims

1. A server computer for predicting the incidence of a disease or disorder, including:

at least one processor and associated memory; and

a communication interface configured to receive a plurality of structured symptom datasets generated by an on-line symptom checking diagnostic tool (OLSC-DT) in response to a corresponding plurality of symptom queries initiated by one or more persons inquiring about symptoms of illness from at least one user device over a distributed computer network, the plurality of structured symptom datasets received prior to a change in public interaction with the distributed computer network;

wherein the at least one process is configured to process the plurality of structured symptom datasets using a dataset algorithm

to produce a multivariate dataset representing relevant symptom counts with geographical and temporal characteristics;

wherein the at least one processor is configured to process the multivariate dataset using a feature construction algorithm to create orthogonal symbols representative of at least one illness;

wherein the at least one processor is configured to create a time series model for the at least one illness based on the orthogonal symbols and historical illness incidence data associated with the at least one illness, the historical illness incidence data having been generated by at least one sentinel provider and being accessible to the at least one processor via the communication interface;

wherein the at least one processor is configured to apply the time series model to new illness incidence data associated with the at least one illness to predict a future incidence of the at least one illness, the new illness incidence data having been generated by one or more sentinel providers and being accessible to the at least one processor via the communication interface; and

wherein the at least one processor is configured to output a geographical and temporal alert of the at least one illness in response to the application of the time series model to the new illness incidence data.

2. The server computer according to claim 1, wherein the historical illness data includes past weekly illness incidence data.

3. The server computer according to claim 1, wherein, in conjunction with using the feature construction algorithm, the at least one processor is configured to perform a principle component analysis (PCA).

4. The server computer according to claim 1, wherein the OLSC-DT comprises a WebMD Symptom Checker.

16. The server computer according to claim 1, wherein the at least one processor is configured to receive filter selections from a supervisor device via the communication interface, the filter selections being associated with control of the at least one processor in conjunction with at least one of the processing of the plurality of structured symptom datasets, the processing of the multivariate dataset, and the creating of the time series model.

18. The server computer according to claim 1, wherein the change in public interaction with the distributed computer network corresponds to user interaction with the distributed computer network following public knowledge of the illness.

5. A method for predicting the incidence of a disease or disorder, the method comprising:

receiving a plurality of structured symptom datasets at a server computer, the plurality of structured symptom datasets having been generated by an on-line symptom checking diagnostic tool (OLSC-DT) in response to a corresponding plurality of symptom queries initiated by one or more persons inquiring about symptoms of illness from at least one user device over a distributed computer network, the plurality of structured symptom datasets received prior to a change in public interaction with the distributed computer network;

processing the plurality of structured symptom datasets at the server computer using a dataset algorithm to produce a multivariate dataset representing medically-relevant symptom counts with geographical and temporal characteristics;

processing the multivariate dataset at the server computer using a feature construction algorithm to create orthogonal symbols representative of at least one illness;

creating a time series model for the at least one illness at the server computer based on the orthogonal symbols and historical illness incidence data associated with the at least one illness, the historical illness incidence data having been generated by at least one sentinel provider and being accessible to the server computer; and

applying the time series model to new illness incidence data associated with the at least one illness at the server computer to predict a future incidence of the at least one illness, the new illness incidence data having been generated by one or more sentinel providers and being accessible to the server computer; and

outputting a geographical and temporal alert of the at least one illness in response to the application of the time series model to the new illness incidence data.

6. The method according to claim 5, further comprising performing a principle component analysis (PCA) in conjunction with using the feature construction algorithm.

7. The method according to claim 5, further including receiving filter selections at the server computer from a supervisor device, the filter selections being associated with control of the server computer in conjunction with at least one of the processing of the plurality of structured symptom datasets, the processing of the multivariate dataset, and the creating of the time series model.

8. The method according to claim 5, further including adjusting a time step lag in the time series model prior to predicting the future incidence of the at least one illness to account for reporting lags in the historical illness incidence data.

9. The method according to claim 8, further including:

filtering the medically-relevant symptom count to create subsets of the multivariate dataset based on previously indicated symptoms associated with at least one previously indicated illness;

wherein the processing of the multivariate dataset is performed on the subsets of the multivariate dataset.

10. The method according to claim 8, wherein the plurality of structured symptom datasets include a plurality of WebMD Symptom Checker structured datasets.

11. The method according to claim 5, wherein epidemiological modeling is used to produce a simulated ill population and representative OLSC-DT queries are produced for the simulated ill population.

13. The method of claim 5, wherein the chosen time interval is a weekly time interval.

14. The method of claim 5, wherein the producing comprises:

performing principal component analysis (PCA) on the medically-relevant symptom counts includes symptom counts for;

(1) Body aches, (2) chills, (3) cough, (4) diarrhea, (5) difficulty breathing through nose, (6) fatigue, (7) fever, (8) headache, (9) headache, (10) joint ache, (11) nausea or vomiting, (12) runny nose, (13) shaking chills, and (14) sore throat.

15. The method of claim 5, wherein the processing associated with the feature construction algorithm comprises:

performing principal component analysis (PCA) on the medically-relevant symptom counts of the multivariate dataset representing counts of symptoms associated with the plurality of structured symptom datasets.

17. The method according to claim 5, wherein the plurality of structured symptom datasets include a plurality of on-line symptom checking diagnostic tool datasets.

19. The method according to claim 5, wherein the change in public interaction with the distributed computer network corresponds to user interaction with the distributed computer network following public knowledge of the illness.

20. The method according to claim 11, wherein the plurality of structured symptom datasets received at the server computer generated by the DLSC-DT are in response to the representative OLSC-DT queries produced for the simulated ill population.

12. A non-transitory computer readable medium storing instructions that, when executed by a processor, cause a processor-controlled server computer to perform a method for predicting the incidence of a disease or disorder, the method comprising:

receiving a plurality of structured symptom datasets at a server computer, the plurality of structured symptom datasets having been generated by an on-line symptom checking diagnostic tool (OLSC-DT) in response to a corresponding plurality of symptom queries initiated by one or more person inquiring about symptoms of illness from at least one user device over a distributed computer network, the plurality of structured symptom datasets received prior to a change in public interaction with the distributed computer network;

processing the plurality of structured symptom datasets at the server computer using a dataset algorithm to produce a multivariate dataset representing medically-relevant symptom counts with geographical and temporal characteristics;

processing the multivariate dataset at the server computer using a feature construction algorithm to create orthogonal symbols representative of at least one illness;

creating a time series model for the at least one illness at the server computer based on the orthogonal symbols and historical illness incidence data associated with the at least one illness, the historical illness incidence data having been generated by at least one sentinel provider and being accessible to the server computer;

applying the time series model to new illness incidence data associated with the at least one illness at the server computer to predict a future incidence of the at least one the new illness incidence data having been generated by one or more sentinel providers and being accessible to the server computer; and

outputting a geographical and temporal alert of the at least one illness in response to the application of the time series model to the new illness incidence data.