Capacity-aware discharge policies reduce readmission learning bias
Original title: Learning through Capacity Awareness to Minimize Hospital Readmissions
Simulations find discharge-up-to policies cut learning error 24%–25% versus threshold-on-risk policies.
The paper extends a queueing-based Markov decision process for hospital discharge so that periodically retrained readmission predictors do not lock in censoring bias. Cost-optimal capacity-aware policies vary length of stay, which the authors say yields consistent risk estimates without explicit fairness constraints. In synthetic experiments and simulations calibrated to a partner hospital, discharge-up-to policies reduce learning error versus threshold-on-risk policies.
Unconfirmed: Results come from synthetic experiments and calibrated simulations, not a live deployment.
- Authors report a 24%–25% drop in learning error versus status-quo thresholding.
- Adjusted readmission cost disparities fall by 38%–63% in their simulations.
- They prove retraining under discharge-up-to policies converges when the model is well specified.
- The partner hospital is not named.
From the original article
1 min read at arxiv.orgPredictive algorithms are increasingly used to guide hospital discharge decisions, yet these models can exhibit censoring bias for vulnerable patients due to insufficient historical data or past discharge decisions.
By Lydia T. Liu · arxiv.org
A short excerpt; the article and its images belong to the publisher. Summary, key facts and assessment above are AI-written from the article and can be wrong.