Dashboard Overview
TM Swing Forecast
Upgrades
Downgrades
TM Swing Forecasting Methodology
Meal order patterns across texture tiers (Pureed PU4, Minced & Moist MM5, Soft & Bite-Sized SB6, Easy Chew EC7, Regular REG7) were compiled from hospital electronic meal ordering systems (eMOS) across daily Breakfast, Lunch, and Dinner services. Daily upgrade and downgrade swing pressure is computed from baseline-normalized proportional transition shifts against a 7-day rolling equilibrium, forecasted via a regularised Patch-ARX ridge model accounting for speech-therapy review cycles and Singapore public holidays (average MAPE: 3.40%).
Daily BOR Forecast
Actual BOR
Predicted BOR
Daily BOR Forecasting Methodology
Daily Bed Occupancy Rate (BOR) records were extracted from public hospital bed utilisation figures published by the Ministry of Health (MOH), Singapore. The forecast is estimated using a seasonal ridge regression model with an autoregressive lag patch framework that incorporates day-of-week discharge and admission rhythms alongside gazetted Singapore public holidays. The current model produces an out-of-sample Mean Absolute Percentage Error (MAPE) of 2.42%.
Monthly Admission (WH)
Actual WH
Predicted WH
Monthly Admission Forecasting Methodology
Monthly inpatient admission statistics for Woodlands Health (WH) were sourced from official public healthcare datasets made available by the Singapore Government under the Singapore Open Data Licence on data.gov.sg. The series is fitted with a multi-lag seasonal ridge model incorporating rolling 3-month autoregressive components, calendar month-length variations, and annual cyclical harmonics. The model delivers an out-of-sample Mean Absolute Percentage Error (MAPE) of 1.22%.