Lowering Barriers to AI Adoption in Regional Hospitals: Predicting Patient Volumes from Minimal Data

Background: Artificial intelligence (AI) is increasingly promoted as a tool to enhance hospital efficiency and patient care. Yet, the adoption of AI in regional hospitals remains limited, often due to data scarcity, high implementation costs, and concerns regarding compliance with data protection laws. This study investigates how predictive analytics based on minimal datasets—limited to admission and discharge timestamps, enriched with contextual public data—can already provide actionable insights for hospital operations. Methods: Using routine hospital data, we aggregated daily admissions, discharges, and inpatient loads, and combined them with external features such as weather and public holidays. While 30 years of data were available, we demonstrate that a training window of only the most recent five years is sufficient to achieve high predictive accuracy. Random Forest models were applied to forecast patient numbers, with performance assessed via mean absolute error (MAE), mean absolute percentage error (MAPE), and root mean squared error (RMSE). The models were designed with a focus on pragmatic AI adoption: simple to implement, explainable, and fully compliant with GDPR through inherent anonymization. Results: Our models achieved high predictive accuracy, capturing both weekly cycles and seasonal fluctuations. Daily inpatient forecasts reached a MAPE of 2.4%, corresponding to an average error of only 10 patients. These results demonstrate that even low-complexity AI can provide reliable decision support for staffing and resource allocation, reducing the risk of overcrowding and improving care delivery. Conclusions: Our findings show that low-complexity, data-efficient AI can provide robust forecasts with minimal inputs, lowering barriers to adoption in regional hospitals while maintaining strong compliance with data protection frameworks.This enables immediate improvements in operational planning, reduces overcrowding risks, and supports care delivery under increasing system pressures.