Generative Modeling of Clinical Time Series via Latent Stochastic Differential Equations

2025-11-20

AI Summary

To address challenges in electronic health records—including irregular sampling, nonlinear physiological dynamics, and measurement uncertainty—this paper proposes a generative modeling framework based on latent-variable neural stochastic differential equations (Neural SDEs). The method jointly models the stochastic evolution of underlying disease dynamics and observation noise via a modality-specific emission model and variational inference, enabling state estimation, parameter learning, and uncertainty quantification for irregular time-series data. Its key innovation lies in embedding stochastic differential equations directly into the latent generative process, explicitly capturing intrinsic physiological randomness and observational uncertainty. Evaluated on simulated lung cancer pharmacokinetics and real-world ICU multivariate physiological signal forecasting, the approach significantly outperforms ODE- and LSTM-based baselines in both predictive accuracy and calibration. This work establishes a new paradigm for interpretable, trustworthy clinical decision support.

Technology Category

• Machine Learning: Calibration & Uncertainty Quantification
• Reasoning under Uncertainty: Stochastic Optimization
• Intelligent Robots: State Estimation

Application Category

• User Modeling, Personalization and Recommendation: Explainable and interpretable methods for personalization
• Search and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for ranking
• Economics, Online Markets and Human Computation: Economic ramifications for generative AI infrastructure and applications

Abstract

Clinical time series data from electronic health records and medical registries offer unprecedented opportunities to understand patient trajectories and inform medical decision-making. However, leveraging such data presents significant challenges due to irregular sampling, complex latent physiology, and inherent uncertainties in both measurements and disease progression. To address these challenges, we propose a generative modeling framework based on latent neural stochastic differential equations (SDEs) that views clinical time series as discrete-time partial observations of an underlying controlled stochastic dynamical system. Our approach models latent dynamics via neural SDEs with modality-dependent emission models, while performing state estimation and parameter learning through variational inference. This formulation naturally handles irregularly sampled observations, learns complex non-linear interactions, and captures the stochasticity of disease progression and measurement noise within a unified scalable probabilistic framework. We validate the framework on two complementary tasks: (i) individual treatment effect estimation using a simulated pharmacokinetic-pharmacodynamic (PKPD) model of lung cancer, and (ii) probabilistic forecasting of physiological signals using real-world intensive care unit (ICU) data from 12,000 patients. Results show that our framework outperforms ordinary differential equation and long short-term memory baseline models in accuracy and uncertainty estimation. These results highlight its potential for enabling precise, uncertainty-aware predictions to support clinical decision-making.

Problem

• Modeling irregularly sampled clinical time series data from medical records
• Capturing complex latent physiology and inherent uncertainties in disease progression
• Generating realistic patient trajectories for treatment effect estimation and forecasting

Innovation

• Latent neural SDEs model clinical time series
• Modality-dependent emission models handle irregular sampling
• Variational inference enables state estimation and learning