Abstract
<title>Abstract</title> <p> <bold>Background:</bold> In stroke care, the challenge is not just determining eligibility for Post-Acute Care (PAC), but prioritizing transfer-planning when resources like beds and staffing are limited. <bold>Objective:</bold> To clarify whether routinely available transfer-time variables can be used to identify PAC-eligible stroke patients with a higher probability of marked functional improvement during the PAC episode, and to translate this information into an interpretable, threshold-based tool for prioritizing early transfer-planning activities. <bold>Methods:</bold> We retrospectively analyzed stroke patients in a PAC program. The outcome was marked functional improvement during PAC. A multivariable logistic regression model included age, Barthel Index (BI), previous stroke, acute-care length of stay (LOS), and Functional Oral Intake Scale (FOIS). Performance was assessed via discrimination, calibration, and decision curve analysis (DCA). <bold>Results:</bold> Younger age, no previous stroke, and higher BI at transfer were independently associated with greater odds of marked improvement. The model demonstrated acceptable discrimination and calibration. DCA indicated clinical utility across a practical threshold range, with maximal incremental net benefit near 0.51. Visualization analyses revealed that BI acts as the primary stratifier for recovery readiness, while FOIS provides crucial refinement for patients near the DCA-supported decision boundary. <bold>Conclusion:</bold> Among PAC-eligible patients, routinely available transfer-time data supports an interpretable tool for prioritizing coordination. While BI provides the main structural signal for prioritization, FOIS adds clinical value for borderline cases near the operational planning threshold. </p>