Demand Forecasting

Concept

Theoretically, Demand Forecasting is the quantitative process of estimating future demand for healthcare services, pharmaceuticals, medical devices, or healthcare resources using historical data, statistical models, and predictive techniques. It supports planning under uncertainty by estimating future utilisation, consumption, or patient volumes, allowing organisations to allocate resources efficiently and evaluate future healthcare needs.

Mathematically, demand forecasting is based on statistical estimation, probability theory, regression analysis, and time-series modelling. Depending on the forecasting horizon and available data, recognised mathematical approaches include moving averages, exponential smoothing, autoregressive integrated moving average (ARIMA) models, regression models, and machine learning methods. Forecast accuracy is commonly assessed using established error metrics.

In practice, demand forecasting is widely applied throughout health economics and healthcare management to predict medicine utilisation, hospital admissions, outpatient activity, vaccination demand, workforce requirements, healthcare expenditure, and disease burden. Forecasts inform budget impact analyses, procurement planning, health system capacity modelling, and economic evaluations.

Purpose

Demand forecasting estimates future healthcare demand using mathematical and statistical models to support planning, budgeting, procurement, capacity management, and policy decision-making while reducing uncertainty in resource allocation.

Mathematical Formulae

Primary Formula

D???? = f(X?, X???, ? , X?)

Where:

• D???? = forecast demand
• h = forecast horizon
• X = historical observations
• f( ) = forecasting model

Supporting Formulae

Moving Average

MA? = (1/n) ? �D?

Simple Exponential Smoothing

F??? = �D? + (1 ? �)F?

Where:

• � = smoothing constant
• D? = observed demand
• F? = previous forecast

Linear Regression Forecast

Y = ?? + ??X + �

Forecast Error

Error = Actual ? Forecast

Mean Absolute Error (MAE)

MAE = (1/n) �|Actual? ? Forecast?|

Root Mean Square Error (RMSE)

RMSE = �[(1/n) �(Actual? ? Forecast?)�]

Mean Absolute Percentage Error (MAPE)

MAPE = (100/n) �|(Actual? ? Forecast?)/Actual?|

Related Mathematical Methods

• Time Series Analysis
• ARIMA Models
• Exponential Smoothing
• Linear Regression
• Multiple Regression
• Seasonal Decomposition
• State Space Models
• Bayesian Forecasting
• Monte Carlo Simulation
• Machine Learning Prediction Models

Example

A hospital records monthly emergency department visits over five years and applies exponential smoothing with � = 0.30.

Current month's demand:

D? = 4,250 patients

Previous forecast:

F? = 4,180 patients

Forecast for next month:

F??? = 0.30 ? 4,250 + 0.70 ? 4,180 = 4,201 patients

Hospital managers use this forecast to schedule staffing, estimate bed occupancy, and plan medical supply requirements.

Excel Implementation

Previous Forecast

=B2

Observed Demand

=C2

Smoothing Constant

=$F$1

Next Forecast

=($F$1*C2)+((1-$F$1)*B2)

Forecast Error

=C2-D2

MAE

=AVERAGE(ABS(E2:E25))

RMSE

=SQRT(AVERAGE(E2:E25^2))

MAPE

=AVERAGE(ABS(E2:E25/C2:C25))*100

VBA (Optional)

Demand forecasting models can be automated in VBA by:

• Performing rolling forecasts
• Updating exponential smoothing calculations
• Estimating regression coefficients
• Calculating forecast accuracy metrics
• Producing forecasting dashboards
• Running scenario analyses

Sources

Hyndman RJ, Athanasopoulos G. Forecasting: Principles and Practice.

Box GEP, Jenkins GM, Reinsel GC, Ljung GM. Time Series Analysis: Forecasting and Control.

Makridakis S, Wheelwright SC, Hyndman RJ. Forecasting Methods and Applications.

Silver EA, Pyke DF, Peterson R. Inventory Management and Production Planning and Scheduling.

Briggs AH, Claxton K, Sculpher MJ. Decision Modelling for Health Economic Evaluation.