Search papers, labs, and topics across Lattice.
The paper introduces MedCertAIn, a framework for improving predictive uncertainty in multimodal clinical risk prediction models. It uses data-driven priors over neural network parameters, informed by cross-modal similarity in self-supervised latent representations and modality-specific data corruptions. Experiments on MIMIC-IV and MIMIC-CXR datasets demonstrate that MedCertAIn significantly enhances both predictive performance and uncertainty quantification compared to deterministic and Bayesian baselines.
Data-driven priors, leveraging cross-modal similarity and modality-specific corruptions, can substantially improve both the accuracy and reliability of uncertainty estimates in multimodal clinical risk prediction.
Safe predictions are a crucial requirement for integrating predictive models into clinical decision support systems. One approach for ensuring trustworthiness is to enable models' ability to express their uncertainty about individual predictions. However, current machine learning models frequently lack reliable uncertainty estimation, hindering real-world deployment. This is further observed in multimodal settings, where the goal is to enable effective information fusion. In this work, we propose $\texttt{MedCertAIn}$, a predictive uncertainty framework that leverages multimodal clinical data for in-hospital risk prediction to improve model performance and reliability. We design data-driven priors over neural network parameters using a hybrid strategy that considers cross-modal similarity in self-supervised latent representations and modality-specific data corruptions. We train and evaluate the models with such priors using clinical time-series and chest X-ray images from the publicly-available datasets MIMIC-IV and MIMIC-CXR. Our results show that $\texttt{MedCertAIn}$ significantly improves predictive performance and uncertainty quantification compared to state-of-the-art deterministic baselines and alternative Bayesian methods. These findings highlight the promise of data-driven priors in advancing robust, uncertainty-aware AI tools for high-stakes clinical applications.
