Academic Awards 2025 booklet

29 Forecasting the Outpatient Demand to Improve Bernhoven’s Session Planning Hospitals are under increasing pressure due to rising care demand and a growing shortage of medical staff. At Bernhoven, this is reflected in among others long waiting times at outpatient clinics. To allocate physicians efficiently, outpatient sessions (fixed time blocks for appointments) must be scheduled three months in advance. However, Bernhoven’s current manual, supply- driven approach fails to account for monthly demand variation, risking care chain build-ups and inefficient resource use. In my thesis, I developed a forecasting model to predict monthly demand at Bernhoven’s cardiology and neurology outpatient clinics. The model differentiates between appointment types: new (NP), emergency (SP), and short- and long-term returning (CP, TC) patients. For each category, I tested various time series methods (e.g., ARIMA, exponential smoothing) and multiple linear regression models with explanatory variables and appointment type interactions. By combining the best-performing methods per appointment type, I created an accurate total session time forecast three months ahead (error <5.5%). This enables planners to identify peak periods and allocate resources accordingly. The forecasts per appointment type also help distribute session time across the different types. This demand and data-driven approach allows Bernhoven to proactively manage outpatient capacity, reduce delays, and contribute to a more accessible and affordable healthcare system. Figure 1: Interaction appointment types at the neurology outpatient clinic, used to derive appointment interaction regressors for short-term returning (CP, TC) patients in the multiple linear regression models. Figure 2: Line plot actual and predicted total session time cardiolgy outpatient clinic. Figure 3: Line plot actual and predicted total session time neurology outpatient clinic.