Learning to Select the Best Forecasting Tasks for Clinical Outcome Prediction
This work addresses clinical risk prediction for healthcare applications, presenting an incremental improvement in representation learning.
The paper tackles the problem of clinical outcome prediction by meta-learning a self-supervised forecasting task to optimize patient representations, achieving better performance with low resources compared to direct supervised learning and pretraining methods on the MIMIC-III dataset.
We propose to meta-learn an a self-supervised patient trajectory forecast learning rule by meta-training on a meta-objective that directly optimizes the utility of the patient representation over the subsequent clinical outcome prediction. This meta-objective directly targets the usefulness of a representation generated from unlabeled clinical measurement forecast for later supervised tasks. The meta-learned can then be directly used in target risk prediction, and the limited available samples can be used for further fine-tuning the model performance. The effectiveness of our approach is tested on a real open source patient EHR dataset MIMIC-III. We are able to demonstrate that our attention-based patient state representation approach can achieve much better performance for predicting target risk with low resources comparing with both direct supervised learning and pretraining with all-observation trajectory forecast.