Transferring healthcare risk prediction models between Medicaid populations: a transfer learning evaluation
Sanjay Basu, Sadiq Y Patel, Parth Sheth, Bhairavi Muralidharan, Namrata Elamaran, Aakriti Kinra, Rajaie Batniji · 2026
Read Transferring healthcare risk prediction models between Medicaid populations: a transfer learning evaluation by Sanjay Basu, Sadiq Y Patel, Parth Sheth, Bhairavi Muralidharan, Namrata Elamaran, Aakriti Kinra, Rajaie Batniji (2026) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/PMC13354197.txt.
Description
Sanjay Basu leads a 2026 transfer learning test of Medicaid risk prediction models across states. Models trained in one state face shifted populations, policies, and care patterns in another. Acute care use was predicted for high-risk care management enrollees in Washington, with 20,744 people, and Virginia, with 28,901 people. Outcome rates differed sharply, at 9.4 percent in Washington versus 25.6 percent in the Virginia group.
Nine approaches were compared, from plain logistic models to neural and ensemble methods. A meta-ensemble scored best on Virginia data, with discrimination of 0.728 and the best calibration of any method tested. Plain source transfer nearly matched it, at 0.725 on the same held-out Virginia data set. Target-only logistic lagged far behind, at 0.628 in the same head-to-head comparison test. Recalibration helped all models, and fairness gaps shrank with target adaptation though disparities persisted.
Catalog record
| CREATOR | Sanjay Basu, Sadiq Y Patel, Parth Sheth, Bhairavi Muralidharan, Namrata Elamaran, Aakriti Kinra, Rajaie Batniji |
|---|---|
| YEAR | 2026 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 49840 |
| LICENSE | https://creativecommons.org/licenses/by/4.0 |
| RIGHTS | CC BY |
| SOURCE | Europe PMC |