Driving Intelligent Decisions in Healthcare: Integrating Actionable Insights, Functional Trajectories, and Transforming System Resilience

Abstract

Delayed hospital discharge, often recorded as Alternate Level of Care, remains a persistent barrier to safe and timely transitions for older adults. This thesis integrates three studies that move from system diagnosis to temporal analysis of functional change to interpretable prediction that can inform planning. Article 1 is an in-depth scoping review of 23 systematic reviews and more than 700 studies. It shows that discharge delays arise from structural and operational gaps in information flow, coordination, and decision rights. It proposes a continuous process improvement model that treats discharge as an iterative cycle of planning, measurement, learning, and adaptation, which sets requirements for later analyses. Article 2 analyzes 878,000 longitudinal observations from Veterans Affairs long term care. Using survival and count models, it quantifies the timing and recurrence of recovery and decline across functional domains. Results show that change is heterogeneous and domain specific, with clear differences by sex and limited added value for chronological age. These findings justify measuring early and late improvement and motivate simple, transparent profiles of recovery. Article 3 applies these ideas to Ontario data from the Institute for Clinical Evaluative Sciences, more than 1.8 million episodes from 2004 to 2023. It engineers early, late, and total gains and constructs rule based Recovery Archetypes, then trains a gradient boosted model to explain ALC duration. Explanations highlight locomotion rate of gain, bathing and shower transfer improvements, and late mobility gains as leading drivers. The framework supports episode-level risk stratification, earlier intervention, capacity planning, and evaluation of policy periods. Together, the studies contribute a system map and process model, a temporal measurement strategy for functional data, and a transparent predictive tool that turns routine records into decision-ready insight for safer and more responsive discharge planning