SuperMAN: Interpretable and Expressive Networks over Temporally Sparse Heterogeneous Data

2025/05/25 by Maya Bechler-Speicher, Zerio, Andrea, Andrea Zerio +14 · 1 voice
Computer Science · #Advanced Graph Neural Networks #Bayesian Modeling and Causal Inference #Domain (mathematical analysis) #Domain Adaptation and Few-Shot Learning #Domain knowledge #ENCODE #Event (particle physics) #FOS: Computer and information sciences #Interpretability #Machine Learning (cs.LG) #Prior probability #SIGNAL (programming language) #Synthetic data #cs.LG

paper · pdf · doi:10.48550/arxiv.2505.19193

published in arXiv (Cornell University) (Cornell University)

openalex publication_date 2025/05/25 · arxiv published 2025/05/25 · openalex created_date 2025/10/10 · arxiv updated 2026/02/28 · openalex updated_date 2026/08/06

Abstract

Real-world temporal data often consists of multiple signal types recorded at irregular, asynchronous intervals. For instance, in the medical domain, different types of blood tests can be measured at different times and frequencies, resulting in fragmented and unevenly scattered temporal data. Similar issues of irregular sampling occur in other domains, such as the monitoring of large systems using event log files. Effectively learning from such data requires handling sets of temporal sparse and heterogeneous signals. In this work, we propose Super Mixing Additive Networks (SuperMAN), a novel and interpretable-by-design framework for learning directly from such heterogeneous signals, by modeling them as sets of implicit graphs. SuperMAN provides diverse interpretability capabilities, including node-level, graph-level, and subset-level importance, and enables practitioners to trade finer-grained interpretability for greater expressivity when domain priors are available. SuperMAN achieves state-of-the-art performance in real-world high-stakes tasks, including predicting Crohn's disease onset and hospital length of stay from routine blood test measurements and detecting fake news. Furthermore, we demonstrate how SuperMAN's interpretability properties assist in revealing disease development phase transitions and provide crucial insights in the healthcare domain.

Citations

• The Interpretable and Effective Graph Neural Additive Networks
• Beyond Random Missingness: Clinically Rethinking for Healthcare Time Series Imputation
• DuETT: Dual Event Time Transformer for Electronic Health Records
• TimesNet: Temporal 2D-Variation Modeling for General Time Series Analysis
• Graph neural networks for materials science and chemistry
• Learning to Reconstruct Missing Data from Spatiotemporal Graphs with Sparse Observations
• SAITS: Self-attention-based imputation for time series
• Filling the G_ap_s: Multivariate Time Series Imputation by Graph Neural Networks
• Self-Supervised Transformer for Sparse and Irregularly Sampled Multivariate Clinical Time-Series
• Highly accurate protein structure prediction with AlphaFold
• CSDI: Conditional Score-based Diffusion Models for Probabilistic Time Series Imputation
• How Attentive are Graph Attention Networks?
• Dynamic Gaussian Mixture based Deep Generative Model For Robust Forecasting on Sparse Multivariate Time Series
• Handling Missing Data with Graph Representation Learning
• Multi-Time Attention Networks for Irregularly Sampled Time Series
• Connecting the Dots: Multivariate Time Series Forecasting with Graph Neural Networks
• Connecting the Dots: Multivariate Time Series Forecasting with Graph Neural Networks
• Set Functions for Time Series
• Fake News Detection on Social Media using Geometric Deep Learning
• Weisfeiler and Leman Go Neural: Higher-order Graph Neural Networks
• How Powerful are Graph Neural Networks?
• Neural Ordinary Differential Equations
• BRITS: Bidirectional Recurrent Imputation for Time Series
• Graph Attention Networks
• Attention Is All You Need
• Inductive Representation Learning on Large Graphs
• Neural Message Passing for Quantum Chemistry
• Semi-Supervised Classification with Graph Convolutional Networks
• Recurrent Neural Networks for Multivariate Time Series with Missing Values
• Efficient Estimation of Word Representations in Vector Space

Discussions

• In our paper for the leading machine learning conference ICLR 2026, we introduce SuperMAN – an interpretable AI model that learns from real-world clinical measurement data, including blood tests, to p [bsky, 3 points, 1 comments]

Related

• Graph Mixing Additive Networks
• Identifying Critical Phases for Disease Onset with Sparse Haematological Biomarkers
• Graph Structure Learning from Unlabeled Data for Event Detection
• Time-aware Heterogeneous Graph Transformer with Adaptive Attention Merging for Health Event Prediction
• Graph-Guided Network for Irregularly Sampled Multivariate Time Series