DISCOVERING ADVERSE HEALTH EVENTS VIA BEHAVIORAL DATA

• US 20140379630A1
• Filed: 06/24/2013
• Published: 12/25/2014
• Est. Priority Date: 06/24/2013
• Status: Abandoned Application

• Current Assignee: Microsoft Technology Licensing LLC (Microsoft Corporation)
• Original Assignee: Microsoft Corporation
• Inventors: Horvitz, Eric J., White, Ryen William
• Application Number:
  US13/924,899
• Publication Number:
  US 20140379630A1
• Time in Patent Office: Days
• US Class Current: 707/603
• CPC Class Codes: G06F 16/2465 Query processing support for facilitating data mining operations in structured databases
• CPC Class Codes: G06F 16/9535 Search customisation based on user profiles and personalisation
• CPC Class Codes: G16H 20/10 relating to drugs or medications, e.g. for ensuring correct administration to patients
• CPC Class Codes: G16H 50/20 for computer-aided diagnosis, e.g. based on medical expert systems
• CPC Class Codes: G16H 50/70 for mining of medical data, e.g. analysing previous cases of other patients
• CPC Class Codes: G16H 70/40 relating to drugs, e.g. their side effects or intended usage

Abstract

Aspects of the subject disclosure are directed towards processing search logs and/or other large scale data sources to detect medical related-effects. For example, an anomalous number of queries regarding a particular symptom and a drug may indicate the existence of a previously unknown side-effect of the drug. Side effects of drug interactions may also be found by processing behavioral data such as queries and social network posts. Also described is the generation of symptom spectra data that is processed to detect anomalies and the like in user behavior corresponding to medical related-effects.

Claims

1. A method comprising, processing large-scale behavioral data to identify health-related effects in which a target outcome is unknown, including recognizing signals in the large-scale behavioral data, including detecting anomalous querying patterns, browsing activities, or both, and taking action upon detecting anomalous querying patterns, browsing activities, or both.

2. The method of claim 1 wherein processing the large-scale behavioral data comprises monitoring events related to one or more medications.

3. The method of claim 1 further comprising, determining based upon interaction behavior whether a set of querying patterns or browsing activities, or both, are indicative of exploratory or experiential information gathering.

4. The method of claim 1 wherein processing the large-scale behavioral data comprises monitoring events related to the use of one or more medical devices or procedures.

5. The method of claim 1 wherein taking action comprises computing significance of an interaction between a plurality of health-related events, including by considering changes between users in different groups of interests gathering information across a spectra.

6. The method of claim 1 wherein recognizing the signals in the large-scale behavioral data comprise monitoring querying patterns over time to determine time-related effects.

7. The method of claim 1 wherein processing the data includes constructing a dataset corresponding to of symptom spectra to detect anomalous or unexpected symptom-related data, or both, within a cohort of users.

8. The method of claim 1 wherein processing the data includes an independence analysis including incorporating at least one of:

background information from users, users of particular drug cardinalities, symptoms for which a drug, device or procedure is usually prescribed, or already-known side effects.

9. The method of claim 1 wherein processing the data includes modeling at least one of:

background processes, prior processes, background levels conditioned on specified factors, topical classes of searches, sessions, pattern evidence in search behavior, probability distributions regarding user goals or cohorts, user cohort reasoning, causality among influences seen via search logs, drug side effects over prolonged periods, combined inferred influences of independent identified factors, multi-drug interactions, or influences and dependences among users and informational goals over time.

10. The method of claim 1 wherein processing the data includes correcting for influence of news or trending interests, or both.

11. The method of claim 1 further comprising, predicting future outcomes regarding an appearance or progression of disorders or symptoms.

12. A system comprising, at least one processor and memory configured as an offline data processing subsystem, the subsystem configured to access behavioral data from one or more sources, and to generate spectra data based upon the behavioral data, the spectra data comprising a probability distribution across a set of data computed using different groups of users.

13. The system of claim 12 wherein the offline subsystem includes a join component configured to combine behavioral data from a plurality of sources.

14. The system of claim 12 wherein the offline subsystem includes a filter component configured to filter based upon at least one:

symptoms, classes or drugs.

15. The system of claim 12 including a user interface configured to render a visible representation of medical-related entities and detected interactions of at least some of the entities.

16. The system of claim 12 wherein the different groups of users comprise at least two of:

users who query for information on one medical entity, users who query for information on another medical entity, or users who query for information on two or more medical entities.

17. The system of claim 12 wherein the system includes an offline subsystem that generates representation data based at least in part upon the spectra data, and an online subsystem that provide a user interface for interacting with the representation data.

18. The system of claim 12 wherein the spectra data corresponds to symptoms or conditions, or both, related to at least one of:

one or more drugs, one or more medical devices or one or more medical procedures.

19. One or more machine-readable storage media or logic having executable instructions, which when executed perform steps, comprising, generating a representation of behavioral data with respect to large scale sets of medical entity information, and processing the representation to recognize health-related effects of one or more medical-related entities based upon statistical analysis, in which a target outcome is unknown.

20. The one or more machine-readable storage media or logic of claim 19 wherein generating the representation of the behavioral data comprises generating a spectra.