Infrastructure for Capacity vs. Demand Forecasting

Demand forecasting requires documented definitions of capacity metrics—what constitutes a 'session,' how RVUs translate to visit capacity, and what third-next-available threshold triggers action. Scheduling templates by provider and appointment type are documented and patient access policies are explicit. However, the forecasting rules themselves—how to weight seasonal patterns, which market indicators signal demand growth, how to translate waitlist length to capacity gap—are not documented. The AI can apply statistical models but lacks documented business rules to contextualize recommendations.

Demand forecasting requires systematic capture of appointment volume by specialty, site, and appointment type over time, with consistent timestamps enabling trend analysis. EHR/PM systems capture all scheduling events through required workflow fields—check-in, cancellation, no-show—producing the historical time-series data needed for seasonal pattern detection. Third-next-available tracking and waitlist length require regular systematic measurement through defined reporting templates. This ongoing, structured capture provides the forecasting model's training and input dataset.

Capacity vs. demand forecasting requires consistent schema: appointment volumes by specialty code, site identifier, appointment type category, and provider productivity metrics (RVUs, visit counts) defined uniformly across the system. The baseline confirms appointment types are categorized and provider schedules are templated. This schema allows the model to aggregate demand signals across sites and compare against structured capacity metrics—producing forecasts by specialty and timeframe rather than system-wide averages.

Demand forecasting operates primarily on historical data exports from EHR/PM reporting modules—appointment volume trends, productivity metrics, waitlist data—which are accessible through scheduling staff interfaces and reporting tools. Real-time programmatic API access to schedule templates is limited, as EHR vendors don't prioritize scheduling APIs. The forecasting model runs on scheduled data extracts rather than live queries, which is sufficient for periodic demand planning cycles but prevents intra-day capacity adjustments.

Demand forecasting models require periodic recalibration as provider productivity standards change, new services launch, or market conditions shift. Provider schedules are updated regularly for blocks and vacations, keeping the capacity input reasonably current. However, the forecasting model's seasonal baselines and trend weights are not systematically reviewed—they update only when someone recognizes the model is producing unreasonable projections. This is sufficient for annual planning cycles but limits responsiveness to mid-year demand shifts.

Capacity vs. demand forecasting requires API-based connections integrating the EHR scheduling system (historical volume), provider productivity reporting (RVU data), and access metrics tracking (third-next-available, waitlist). The forecast outputs must flow back to schedule management tools and workforce planning systems to enable actionable capacity adjustments. API connections across these systems allow the model to assemble the full capacity-demand picture and publish recommendations to the platforms where scheduling and staffing decisions are made.