Identity Is Key: How to Unlock Big Data and Analyze Populations

Transcription

1 Identity Is Key: How to Unlock Big Data and Analyze Populations Brent Williams

2 Session Objectives Provide the audience with an understanding of how aggregating information from multiple sources (claims, clinical, biographical, etc.) is the key to analyzing populations Discuss how the challenge of aggregating information boils down to the problem of correctly attributing this information to the correct patient Inform the audience that identity information is the key to correctly attributing information to the correct patients, because current matching technologies rely on identity attributes like name, address, and birthdate to match information to patients Go over how a big data approach to identity information enables more successful and accurate information attribution

3 Why is population health important? $ Raise Incentives Lower Costs Population Health Patient Health and Welfare More Timely

4 Population health: the Old World Course of care Diagnosis Care event Correctly Measured Medical event Outcome Incorrectly Measured Minimal analysis can be performed

5 Population health: the Current World Course of care Diagnosis Care event Care event Rx Care event Rx Medical event Outcome More detailed analysis to provide better outcomes in the future

6 Population health: the Emerging World Course of care Diagnosis Care event Care event Rx Care event Rx Medical event Outcome Lab result Radiology Lab result More detailed analysis to provide better outcomes in the future

7 Population health: the Target World Course of care Family info Medical event Outcome Non-medical event Medical device info Employment info Location info More detailed analysis of past and present to provide better outcomes in the future

8 Imagine if my providers had a 360-degree view of me Diagnoses Family Lab results Prescriptions Connected medical devices Radiology Care events Location Non-medical events Employment

9 CLAIMS Diagnoses Prescriptions Care events How do you aggregate and access this data? CLINICAL Lab results Radiology BIOGRAPHIC Family Location Employment Link and exchange patient information across systems and organizations. Enrich patient information with third-party data. OTHER Non-medical events Connected medical devices

10 Gain longitudinal views of patients to better analyze and understand populations CLAIMS Diagnoses Prescriptions Care events CLINICAL Lab results Radiology BIOGRAPHIC Family Location Employment OTHER Non-medical events Connected medical devices Analytics that are: Deeper More complete More accurate

11 Gain more relevant courses of care and more granular outcome measurements Course of Care Outcome Measurements Analytics that are: Deeper More complete More accurate

12 Gain more relevant courses of care and more granular outcome measurements Course of Care Outcome Measurements Analytics that are: Deeper More complete More accurate

13 Apply the analytics in the clinical or aftercare environment 30% higher risk for heart disease New prescription Analytics that are: Deeper More complete More accurate History of poor diet and exercise Dust exposure Lung function test Solicit support of family to develop new regimen

14 Identity data is the key to linking patient records and appending big data CLAIMS Diagnoses Prescriptions Care events CLINICAL Lab results Radiology BIOGRAPHIC Family Location Employment OTHER Non-medical events Connected medical devices Name Address Date of Birth SSN Gender

15 Identity data is a collection of attributes that often change over time Bio graph Name(s), Address, Gender, Birthdate Family graph Parents, Siblings, Spouses, Children Cyber graph Phone, , Social IDs, Device IDs, IP Government graph SSN, National Benefits ID, Drivers Permits, Legal Records Social graph Friends, Roommates, Associations Professional graph University, Degrees, Employers, Coworkers, Prof credentials, Associations Commercial graph Loyalty cards, Purchases Financial graph Fin. Accounts, Homes, Cars, Boats, Mortgages, Loans, Credit

16 30-40% of all medical records have errors in their identity data One person s data... experiences a wide array of errors Time Name change Address change Phone/ change Married Divorced Birth Death... resulting in disparate identity representations across systems 1 Ambiguity Errors Data Gov. Hyphenated name Nicknames Jr/Sr, Twins Default entries Missing data Spelling error Transcription error Homonym error Data quality Privacy controls Formatting 2 3

17 There is a new way to match patient identities together that goes beyond deterministic or probabilistic matching algorithms to make a match despite bad identity data. It s called referential matching and it utilizes big data.

18 Referential matching utilizes big data databases of reference identity data These big data databases aggregate identity data from various sources: Credit header data Telco record identity data Gov t & legal record identity data

19 Two patient records with very different data can still match to each other

20 Two patient records with very different data can still match to each other Match

21 Better matching using a big data approach enables the future of population health Current World Target World Accurate 360-degree view of patient Medical Event Course of Care Outcome Custom course of care Granular outcome measurement Deep population analysis

22 Brent Williams linkedin.com/in/jbrentwilliams