Constructing the Next Generation of Integrated Data Systems Ramesh Raghavan PARTNERING for CHANGE
Disclosures Dr. Ramesh Raghavan has no relevant financial relationships or conflicts of interest to disclose
- Andreas Weigend, former chief scientist, Amazon “Data is the new oil.” - Andreas Weigend, former chief scientist, Amazon
What do we mean by data integration? Physically bring data together in one place (e.g., Extract, Transform, Load – ETL solutions) Physically copy data across various locations (e.g., Integration Platform as a Service) Formerly, Enterprise Application Integration or Enterprise Data Integration Data repositories (e.g., common data storage solutions or “warehouses”) Data virtualization (e.g., “screen scraping” or dashboarding) Use a global schema (what a user needs) + A local schema (the sources from which data are extracted) + A mapping approach
Overview Why integrate data? Selected high performance data integrations Common elements and challenges in data integration
Overview Why integrate data? Selected high performance data integrations Common elements and challenges in data integration
Social Determinants are Key to Health Outcomes
DeBord DG, Carreón T, Lentz TJ, Middendorf PJ, Hoover MD, Schulte PA DeBord DG, Carreón T, Lentz TJ, Middendorf PJ, Hoover MD, Schulte PA. Use of the “Exposome” in the Practice of Epidemiology: A Primer on -Omic Technologies. American journal of epidemiology. 2016;184(4):302-314.
Health Plans Have to Report on These Determinants Contractors’ Obligations Develop relationships with local organizations to implement social determinants interventions (housing support, nutrition classes, exercise equipment) Collaborate on community needs assessments Develop community resource directory Evaluate members’ health related social needs Utilize data to reduce health disparities Share information (health records) with community organizations Evaluation to measure success of social determinants of health interventions Bonus payments for addressing social determinants of health NASHP analysis of accountable health initiatives in 12 states (CA, CO, CT, DE, MA, MI, MN, NY, OR, RI, VT, and WA) that emphasize population health in their models.
Overview Why integrate data? Selected high performance data integrations Common elements and challenges in data integration
Geisinger Health’s Unified Data Architecture (2017 onwards) Shift from large data to big data Create a Hadoop-based platform Build source ingestion pipelines Some data elements are in the enterprise data warehouse (EDW) Some are in free text – analyzed using natural language processing “We are on our way to proving that we can replace the EDW.” Joseph Scopelliti, quoted in Healthcare Informatics (https://www.healthcare-informatics.com/article/analytics/innovator-awards-semi-finalist-team-geisinger-health-s-unified-data-architecture)
Comprehensive Child Welfare Information System 45 CFR 1355.52 (e): Addition of new data exchanges Courts Education Medicaid claims 1355.52 (f): Establish common data exchange standards As of 8/1/2018, 46 states, DC, and PR have declared as CCWIS
NJ’s integrated Population Health Data (iPHD)
High-Touch Data Systems “Alberta and other Canadian provinces function as public health systems. …They are already collecting patient reported outcomes data. It’s more than just collecting the data that exists. It’s acknowledging the data you need but does not exist, and then building a strategic roadmap to fill the gaps.” (Dale Sanders) Such data includes factors like income, language, patient-activation measures, and all the lifestyle behaviors that have an impact on a person’s health such as air and water quality, sexual behaviors, diet, alcohol use, smoking, diet, housing and access to transportation. https://www.healthcatalyst.com/integrating-data-across-systems-of-care-4-perspectives
Overview Why integrate data? Selected high performance data integrations Common elements and challenges in data integration
What Does the Next Generation of Integrated Data Systems Look Like? “Big” data (not large data) Require novel analytics Expand prediction science Integrate social + geospatial ecology Instantiate “post-privacy” world Shift from sciences of outcomes (e.g., medicine), to sciences of populations (e.g., public health), to sciences of contexts (e.g., social work)
Our Future Challenges Technical Common ontologies Source identification Source ingestion Sunk costs in older data systems Data quality (=completeness x validity) Ethical Personal privacy and consent Data control/fair use Use of conditional predictions
Whether or not it draws on new scientific research, technology is a branch of moral philosophy, not of science. Paul Goodman, New Reformation
Ramesh Raghavan ramesh.raghavan@rutgers.edu 848-932-5337 Thank You Ramesh Raghavan ramesh.raghavan@rutgers.edu 848-932-5337 PARTNERING for CHANGE