Variational Disentanglement for Rare Event Modeling

TL;DR

This paper introduces a variational disentanglement method designed to improve rare event prediction in heavily imbalanced healthcare datasets, leveraging latent space behavior and combining additive models with neural nets.

Contribution

It proposes a novel semi-parametric approach that effectively learns from low-prevalence events using a variational framework and a robust prediction model.

Findings

Outperforms existing methods on synthetic and real-world datasets.

Effective in predicting rare events like COVID-19 mortality.

Leverages latent space to extract information from low-prevalence cases.

Abstract

Combining the increasing availability and abundance of healthcare data and the current advances in machine learning methods have created renewed opportunities to improve clinical decision support systems. However, in healthcare risk prediction applications, the proportion of cases with the condition (label) of interest is often very low relative to the available sample size. Though very prevalent in healthcare, such imbalanced classification settings are also common and challenging in many other scenarios. So motivated, we propose a variational disentanglement approach to semi-parametrically learn from rare events in heavily imbalanced classification problems. Specifically, we leverage the imposed extreme-distribution behavior on a latent space to extract information from low-prevalence events, and develop a robust prediction arm that joins the merits of the generalized additive model and isotonic neural nets. Results on synthetic studies and diverse real-world datasets, including mortality prediction on a COVID-19 cohort, demonstrate that the proposed approach outperforms existing alternatives.