Predictive Multimodal Modeling of Diagnoses and Treatments in EHR
This work addresses the challenge of predictive modeling in healthcare for early risk identification and resource optimization, representing an incremental improvement over existing methods.
The paper tackles the problem of early forecasting of diagnoses and treatments in electronic health records using limited information at the start of a patient stay, proposing a multimodal system that fuses clinical notes and tabular events with strategies like cross-modal attention and weighted temporal loss, resulting in enhanced performance that outperforms state-of-the-art systems.
While the ICD code assignment problem has been widely studied, most works have focused on post-discharge document classification. Models for early forecasting of this information could be used for identifying health risks, suggesting effective treatments, or optimizing resource allocation. To address the challenge of predictive modeling using the limited information at the beginning of a patient stay, we propose a multimodal system to fuse clinical notes and tabular events captured in electronic health records. The model integrates pre-trained encoders, feature pooling, and cross-modal attention to learn optimal representations across modalities and balance their presence at every temporal point. Moreover, we present a weighted temporal loss that adjusts its contribution at each point in time. Experiments show that these strategies enhance the early prediction model, outperforming the current state-of-the-art systems.