Capacity Aware Prioritization of Clinical Decision Support Alerts
DOI:
Keywords:
Clinical decision support; alert workload; probability calibration; capacity constraints; operational evaluation; health informatics.
Abstract
Clinical prediction systems can generate more alerts than a review team can process, making available staff time an essential part of evaluation. This paper develops a transparent analytical model for prioritizing a hypothetical routine alert queue under a fixed review budget. Four constructed profiles differ in actionable finding probability, case count, and expected review duration. A probability threshold, risk ranking, findings maximization, and findings maximization with a protected review requirement are compared. With 240 minutes available, risk ranking reviews 20 cases and identifies eight expected actionable findings. Unrestricted optimization reviews 75 cases and identifies 11.75 expected findings, but excludes the highest probability profile. Requiring eight protected reviews from that profile yields 56 reviews and 10.40 expected findings. A threshold applied without a capacity constraint requires 780 minutes and is infeasible. Sensitivity analysis shows that underestimated review times and miscalibrated probabilities can substantially reduce the apparent advantage. All results are calculated from synthetic inputs and do not demonstrate improved clinical outcomes. The contribution is a reproducible operational framework that makes workload, objective choice, and protection rules explicit before any prospective clinical evaluation.
Downloads
Published
Issue
Section
License

This work is licensed under a Creative Commons Attribution-NonCommercial 4.0 International License.

