Towards Multimodal Representation Learning in Paediatric Kidney Disease

2025-11-17

AI Summary

This study addresses the clinical need for early warning of pediatric kidney disease by proposing a multimodal temporal representation learning framework to predict serum creatinine abnormalities within 30 days. Methodologically, it integrates longitudinal laboratory time series (e.g., blood urea nitrogen, electrolytes) with static demographic features into a lightweight recurrent neural network architecture; a novel, simple yet robust temporal encoding module is introduced to ensure clinical interpretability and scalability to heterogeneous data sources. Evaluated on real-world electronic health record data, the model significantly outperforms unimodal baselines, achieving an AUC of 0.82 while demonstrating strong generalizability across institutions and patient subgroups. This work establishes a deployable, multimodal modeling paradigm for dynamic monitoring of pediatric chronic kidney disease and empirically validates the efficacy and clinical utility of temporal modeling in child-specific predictive analytics.

Technology Category

• Machine Learning: Multimodal Learning
• Computer Vision: Multi-modal Vision
• Data Mining & Knowledge Management: Mining of Spatial, Temporal or Spatio-Temporal Data

Application Category

• User Modeling, Personalization and Recommendation: On-Device user modeling, personalization, and recommendation
• Graph Algorithms and Modeling for the Web: Representation, reconstruction, and subgraph or motif discovery in Web-related graphs
• Semantics and Knowledge: Methods to enhance, augment, integrate or synergize semantic models such as knowledge graphs and LLMs

Abstract

Paediatric kidney disease varies widely in its presentation and progression, which calls for continuous monitoring of renal function. Using electronic health records collected between 2019 and 2025 at Great Ormond Street Hospital, a leading UK paediatric hospital, we explored a temporal modelling approach that integrates longitudinal laboratory sequences with demographic information. A recurrent neural model trained on these data was used to predict whether a child would record an abnormal serum creatinine value within the following thirty days. Framed as a pilot study, this work provides an initial demonstration that simple temporal representations can capture useful patterns in routine paediatric data and lays the groundwork for future multimodal extensions using additional clinical signals and more detailed renal outcomes.

Problem

• Predicting abnormal kidney function in children using longitudinal health data
• Integrating laboratory sequences with demographics for temporal modeling
• Developing multimodal representation learning for pediatric renal disease monitoring

Innovation

• Recurrent neural model integrates longitudinal lab data
• Predicts abnormal creatinine values within thirty days
• Uses temporal modeling with demographic information