A Big Data architecture for intra-surgical remaining time predictions

The operating room area is still one of the most expensive sections in the hospital due to the high resource requirements and the diverse uncertainties. However there are few solutions that support monitoring and decision-making in operating room management. But with new data sources and analytical methods of big data research more improvements could be achieved. In this work we utilize surgical phase events recognized in surgical device data to learn prediction models and trigger online predictions for remaining intervention times in operating rooms. To identify the best algorithm for prediction model computation with the existing data, we evaluate a set of regression algorithms. Based on this methods we propose an architecture approach for the integrated processing of real-time data and historic learning data. The evaluation and comparison with related work shows that our prototype is competitive regarding prediction accuracy.