From Ward to Home: Harnessing Agentic AI to Identify Hospital-in-the-Home Patients

Overview

Hospitals are under increasing pressure to manage inpatient capacity while maintaining safe, high-quality care. This research collaboration explores how agentic AI can support earlier transitions from hospital wards to home-based care through Hospital in the Home (HITH) programs.

Unlike conventional predictive models, the proposed agentic AI framework combines structured clinical data with unstructured inputs such as free-text notes to identify patients suitable for HITH. It provides transparent, explainable reasoning behind each recommendation, allowing clinicians to interrogate and adapt the system in real-world practice. Through iterative action research cycles, the team led by PhD researcher Iman Izzati Mohammad Firdaus is co-designing an AI-supported workflow that integrates seamlessly into clinical decision-making.

The project is supported by health service leadership and aims to reduce strain on inpatient resources while improving patient outcomes. By identifying low-risk patients who can safely recover at home, the system helps optimise bed usage, reduce costs, and enhance continuity of care.

This project highlights how next-generation AI can be embedded responsibly into healthcare operations. By enabling clinicians to understand and trust AI recommendations, it supports safer, earlier transitions to home-based care and helps alleviate pressure on hospital resources. More broadly, it demonstrates how agentic AI—designed to reason, not just predict—can be integrated into enterprise workflows to improve decision quality, operational efficiency, and patient outcomes. The result is a scalable framework for deploying transparent, human-centred AI across hospital systems, transforming how care is delivered from ward to home.