UNIFIED HEALTHCARE DATA INFRASTRUCTURE FOR PATIENT FLOW, BED UTILIZATION, CLINICAL OPERATIONS AND HOSPITAL PERFORMANCE ANALYTICS - (HOSPITAL PULSE)

Abstract

Hospitals generate a large volume of operational data every day through registrations, admissions, transfers, discharges, laboratory orders, operation theatre schedules, and billing. This data is usually spread across separate systems for the outpatient desk, wards, laboratory, pharmacy, and finance, and each department sees only its own part of the picture. As a result, questions such as how many beds will be free this evening, why emergency patients wait for admission, or which ward has the longest length of stay are answered slowly and often from memory. This paper presents Hospital Pulse, a unified healthcare data infrastructure that integrates these sources and provides analytics for patient flow, bed utilisation, clinical operations, and hospital performance. The infrastructure is built around an extract, transform, and load pipeline together with change data capture from the hospital information system. Admission, discharge, and transfer events, bed status updates, laboratory and radiology orders, theatre bookings, and billing records are ingested into a raw layer, standardised into a common patient encounter model, and loaded into a clinical operations data warehouse. Patient identifiers are pseudonymised before data enters the analytical layer, and role-based access controls limit what each user can see. On the integrated data, the system computes key performance indicators such as bed occupancy rate, average length of stay, bed turnover, emergency department waiting time, theatre utilisation, and laboratory turnaround time. Predictive models add a forward-looking view. A gradient boosting model estimates each inpatient's remaining length of stay and the probability of discharge within 24 hours, and a timeseries model forecasts daily admissions by department. Together they give a projection of bed availability for the next day, which bed managers can use to plan admissions and transfers. The same data also shows where time is lost during the day, for example between a discharge decision and the release of the bed after billing clearance and cleaning.