Overcoming Uncertain Incompleteness for Robust Multimodal Sequential Diagnosis Prediction via Curriculum Data Erasing Guided Knowledge Distillation
Heejoon Koo
TL;DR
This work tackles robust multimodal sequential diagnosis prediction (SDP) under uncertain missingness in clinical records. It introduces NECHO v2, which modifies the original NECHO to manage varying modality dominance and deploys a systematic knowledge distillation (KD) framework that includes modality-wise contrastive and hierarchical distillation, transformer representation random distillation, MAG distillation, and dual logit distillation, complemented by curriculum data erasing to simulate missing visits. On the MIMIC-III dataset, NECHO v2 achieves state-of-the-art performance across balanced and imbalanced missingness settings, with ablation analyses confirming the contribution of each KD component and the data-erasing strategy. The approach offers a practical, scalable path for robust SDP in real-world incomplete data contexts and is accompanied by code release to facilitate future research and application.
Abstract
In this paper, we present NECHO v2, a novel framework designed to enhance the predictive accuracy of multimodal sequential patient diagnoses under uncertain missing visit sequences, a common challenge in real clinical settings. Firstly, we modify NECHO, designed in a diagnosis code-centric fashion, to handle uncertain modality representation dominance under the imperfect data. Secondly, we develop a systematic knowledge distillation by employing the modified NECHO as both teacher and student. It encompasses a modality-wise contrastive and hierarchical distillation, transformer representation random distillation, along with other distillations to align representations between teacher and student tightly and effectively. We also propose curriculum learning guided random data erasing within sequences during both training and distillation of the teacher to lightly simulate scenario with missing visit information, thereby fostering effective knowledge transfer. As a result, NECHO v2 verifies itself by showing robust superiority in multimodal sequential diagnosis prediction under both balanced and imbalanced incomplete settings on multimodal healthcare data.