Deep continual multitask severity assessment from changing clinical features

Abstract

When developing Machine Learning models to support emergency medical triage, it is important to consider how changes over time in the data distribution can negatively affect the models' performance. The objective of this study was to assess the effectiveness of various Continual Learning pipelines in keeping model performance stable when input features are subject to change over time, including the emergence of new features and the disappearance of existing ones. The model is designed to identify life-threatening situations, calculate its admissible response delay, and determine its institution jurisdiction. We analyzed a total of 1 414 575 events spanning from 2009 to 2019. Our findings demonstrate important performance improvements, up to 7.8% in life-threatening and 14.8% in response delay, in terms of F1-score, when employing deep continual approaches. We noticed that combining fine-tuning and dynamic feature domain updating strategies offers a practical and effective solution for addressing these distributional drifts in medical emergency data.