Interpreting Differentiable Latent States for Healthcare Time-series Data
This work addresses interpretability challenges for deploying advanced machine learning in digital healthcare, which is crucial for clinical applications like disease pattern identification and patient outcome prediction, but it appears incremental as it builds on existing differentiable models.
The paper tackles the problem of limited interpretability in machine learning models for healthcare time-series data by presenting a differentiable algorithm that interprets latent states and predictions, demonstrating its application in identifying a daytime behavioral pattern to predict nocturnal behavior in a real-world dataset.
Machine learning enables extracting clinical insights from large temporal datasets. The applications of such machine learning models include identifying disease patterns and predicting patient outcomes. However, limited interpretability poses challenges for deploying advanced machine learning in digital healthcare. Understanding the meaning of latent states is crucial for interpreting machine learning models, assuming they capture underlying patterns. In this paper, we present a concise algorithm that allows for i) interpreting latent states using highly related input features; ii) interpreting predictions using subsets of input features via latent states; and iii) interpreting changes in latent states over time. The proposed algorithm is feasible for any model that is differentiable. We demonstrate that this approach enables the identification of a daytime behavioral pattern for predicting nocturnal behavior in a real-world healthcare dataset.