Towards Dynamic Feature Acquisition on Medical Time Series by Maximizing Conditional Mutual Information
This addresses cost reduction in medical monitoring and wearables, but it is incremental as it does not yet beat existing static methods.
The paper tackles the problem of dynamically selecting which features to measure in medical time series to reduce costs while maintaining prediction performance, proposing an approach that trains acquisition policies end-to-end using downstream loss. The method outperforms random acquisition and matches an unrestrained budget model but does not surpass a static acquisition strategy.
Knowing which features of a multivariate time series to measure and when is a key task in medicine, wearables, and robotics. Better acquisition policies can reduce costs while maintaining or even improving the performance of downstream predictors. Inspired by the maximization of conditional mutual information, we propose an approach to train acquirers end-to-end using only the downstream loss. We show that our method outperforms random acquisition policy, matches a model with an unrestrained budget, but does not yet overtake a static acquisition strategy. We highlight the assumptions and outline avenues for future work.