MIT GOV/LAB DSSP — Bayes Impact
0 likes48 views
Bayes Impact is a nonprofit that uses data science to monitor health plan compliance and ensure network adequacy. Their solution involves using representative population points and algorithms to measure travel times and distances to healthcare providers. This allows governments to identify areas with insufficient access and collaborate with health plans to improve networks. Their tool has been implemented in California to partner with the Department of Managed Health Care. They aim to expand this work to other states and the federal level to evaluate access for millions of people.
![Bayes-DMHC partnership
13
The [Bayes Impact Network Adequacy] application transforms the way
governments and health care advocates measure and ensure reliable
access to care for millions of people. It enables governments to identify
areas of insufficient access and collaborate with health plans to build
comprehensive networks that serve everyone.](https://image.slidesharecdn.com/bayesimpact-mitgov2flab12-07-2017-171213221906/85/MIT-GOV-LAB-DSSP-Bayes-Impact-13-320.jpg)
MIT GOV/LAB DSSP — Bayes Impact
- 1. 1
- 2. Using Data Science to Monitor Health Plan Compliance Mehdi Jamei, Executive Director, Bayes Impact Dec 7, 2017 - MIT GOV/LAB 2
- 3. TOC 3 ● Bayes Impact, and introduction ● Bayes’ health vision ● What is network adequacy? ● Policy ○ National level ○ State level ● Current solutions ● Our solution ● Product demo ● Expanding the impact ● Government as a platform ● What is next?
- 4. Bayes Impact is a technology nonprofit that builds data-driven social services to empower the underserved for a more equitable future. OUR MISSION
- 5. What do we do? Augment the impact of policy through technology. Policy -> Execution tools -> Expanding the impact 5
- 6. A Health Care system that is affordable, accessible, and high-quality. BAYES’ HEALTH VISION
- 7. ACA→ More comprehensive benefits → Narrow networks → Decreased access & quality :-( 7
- 8. 22M buy marketplace insurance 55M depend on Medicaid 59M live in rural areas 8
- 9. 99
- 10. Policy at national level 1010
- 11. Policy at state level 11 California Massachusetts Michigan
- 12. 12
- 13. Bayes-DMHC partnership 13 The [Bayes Impact Network Adequacy] application transforms the way governments and health care advocates measure and ensure reliable access to care for millions of people. It enables governments to identify areas of insufficient access and collaborate with health plans to build comprehensive networks that serve everyone.
- 14. Our solution - Part 1 14 Goal: To accurately represent the population, and where they live and work. Farthest-First Traversal Algorithm ● Idea: Choose representative points greedily. At each stage, find the single least represented point (measured by distance from any already-selected points) and select it as a representative point ● Data: USPS mail delivery route, EDDM ● Key Features: ○ Deterministic ○ Performant ○ Sequential ○ Configurable
- 15. Our solution - Part 2 15 Goal: A scalable and accurate method to measure travel time and distance standards. - Scale - 101k representative points - 70k PCPs - More than 100 networks in CA - Isodistance and isochrones - Haversine vs drive distance - Traffic and public transit
- 16. Product Demo 16 Part 1: Representative Population Generator (RPG) (link) Part 2: Time and Distance Standards (link)
- 17. Expanding the Impact 17 Goal: Ensuring an adequate access to health care for all. Potential applications 1. More informed interventions 2. In-network provider finder for the public
- 18. Expanding the Impact 18 Health Service Delivery Centers Goal: To study the distribution of federally funded clinics and asses if they are located in the most high-need areas. Key takeaways: ● Uninsured and low-income populations depend on these clinics for care ● Rural areas have least access, and fewest transportation options Access to Health Service Delivery Centers in Southern California
- 19. Expanding the Impact 19 Access to provider’s network for the public Goal: Empower the public to find providers Key takeaways: ● No easy method for the public to access the directory ● The data is not reliable, nor discoverable
- 20. Realization of government as a platform, enables us to 1) bring transparency, and built trust 2) amplify government’s impact and reach.
- 21. What’s next - short term 21 ● Develop the tool for DMHC ● Expand beyond CA ○ Other states ○ Federal government ● Help Gov. in designing meaningful metrics ● Partnership with researchers and policy makers ● Design a tool to empower impactful intervention. What is the change on the ground? ● Build a roadmap, and ask the community to help!
- 22. What’s next - medium term 22 ● Repeat this model in other domains ○ Quality of care, QPP program with CMS ○ Price transparency ○ Beyond Health care
- 23. Policy researchers Linguists Cat lovers UX researchers Gardeners Photographers Data scientists Sailors DevOps Pun enthusiasts Software engineers Pianists Product managers Foodies Designers Writers Physicists Backpackers Bayes Impact team - bayesimpact.org 23
- 24. Questions?! 24