Predicting emergency patient inflow

KickstartAI X Erasmus MC - Time series analysis / statistics

Together with Erasmus MC, we used AI to make their Emergency Department (ED) more data driven. The ED faced challenges with unpredictable patient inflow. We used AI to improve patient flow management and optimize staffing. The key was to forecast short-term patient arrivals and analyze long-term trends.

Staying ahead of surging patient inflow

The hospital’s ED frequently reached maximum capacity (‘going on red’). This was due to unpredictable peaks in patient arrivals and limited treatment capacity. Current forecasting was intuition based without any existing data-driven system in place. We developed an AI solution for real-time actionable predictions, leveraging data to enhance operations and decision-making in the ED  

Goals

1. Develop a tool to forecast near-future patient inflow

2. Help staff to proactively manage patient flow

3. Analyze long-term patient inflow patterns for optimal shift planning

4. Identify trends and patterns in patient arrivals 

Approach

Our AI system combines short-term forecasting with long-term trend analysis for accurate predictions. By forecasting patient arrivals for the next few hours, the system helps staff anticipate sudden surges and prepare for worst-case scenarios. Simultaneously, historical data is analyzed using time series modeling to identify long-term patterns, providing valuable insights for strategic planning and resource allocation.

Key features

• Real-time forecasting of patient inflow

• Long-term analysis of patient inflow trends to optimize staffing and resources

• Integration with Erasmus MC's Integral Capacity Management (ICM) team 
  

Conclusion

The AI-powered solution we developed with the Erasmus MC Emergency Department significantly improves their capacity management. With real-time forecasting and long-term trend analysis, the ED can now make data-driven decisions. It improves their resource allocation and patient flow management.

Results (project in progress)

• Improve ED staff's ability to anticipate and manage patient inflow.

• Reduce the frequency of the ED reaching maximum capacity ("going on red").

• Optimize staff scheduling and resource allocation based on long-term patient inflow patterns.

• Enhance Erasmus MC's expertise in utilizing AI for optimizing healthcare operations.