كنفرانس بين المللی مهندسی صنايع و سيستم ها

صفحه اصلی / دوازدهمین كنفرانس بين المللی مهندسی صنايع و سيستم ها

An Explainable Early-Warning Framework for Hospital Performance Management

نویسندگان :

Zahra Mojaradi¹ Ali Bozorgi-Amiri² Zeinab Sazvar³

1- دانشگاه تهران 2- دانشگاه تهران 3- دانشگاه تهران

کلمات کلیدی :

balanced scorecard،hospital performance،machine learning،explainable AI،early warning،decision support

چکیده :

Abstract-This paper extends a previously published Balanced Scorecard-machine-learning (BSC-ML) hospital performance model [10] by adding a transparent decision-prioritization layer rather than proposing a second predictive model. The baseline study used 28 quarterly observations (2018Q1-2024Q4), 28 input indicators and four composite BSC outputs. The proposed Managerial Intervention Priority Index (MIPI) combines normalized performance deviation, published model-derived importance and consequence criticality. Validation is deliberately multi-layered: nine synthetic stress-test scenarios examine internal logic; five expert ranking profiles provide scenario-level face validation; and an additional archival BSC dataset containing 11 annual observations (1392-1402 Solar Hijri) is used for retrospective real-data back-testing. Eight indicators with usable annual series and monotonic performance direction were included in the retrospective analysis. Historical priority scores ranged from 0.381 to 0.626, with the highest retrospective priority in 1393, followed by 1397 and 1396. These findings support the internal consistency and retrospective discriminative behavior of the framework, but do not establish prospective clinical or managerial effectiveness. Multicenter prospective validation and direct export of raw model-importance coefficients remain necessary before operational deployment.