Post-discharge recovery is a dynamic process in which patient conditions, behaviors, and care decisions evolve after the patient leaves the hospital. During this period, clinical uncertainty increases because physiological abnormalities may fluctuate, symptoms may be self-reported or incomplete, and the effect of treatment or lifestyle guidance may only become visible over time. Existing systems often focus on monitoring or risk prediction, but they provide limited support for reasoning about how a patient’s recovery trajectory may change under alternative care actions. To address this gap, this paper proposes a causal temporal Bayesian reasoning framework for post-discharge recovery management. The framework represents patient recovery as a longitudinal case and uses a two-slice temporal Bayesian network to model the probabilistic evolution of readmission risk and recovery status. Three reasoning functions are developed within the framework. Observational inference updates the patient’s current risk and recovery status from newly available monitoring evidence. Interventional inference compares candidate follow-up actions, such as medication adjustment or lifestyle guidance, by estimating their expected effects on future recovery states. Counterfactual inference is activated after case closure to review whether alternative prior decisions may have produced a more favorable recovery trajectory. The proposed framework is positioned within a broader post-discharge recovery management system that connects patient monitoring, case record updating, provider review, and retrospective learning. An illustrative hypertension recovery case study is used to demonstrate how longitudinal evidence can be processed through the framework to support risk estimation, intervention comparison, and post-case review.

Keywords

Causal reasoning, Post-discharge recovery management, Temporal Bayesian network, Interventional inference, Counterfactual reasoning.