---
title: "Multimodal Empirical Bayes Variational Autoencoders for Joint Longitudinal and Time-to-Event Modeling"
source: "arXiv AI/CL/LG（论文）"
source_tier: "t2"
url: "https://arxiv.org/abs/2607.13984v1"
permalink: "https://feed.astralor.com/items/itm_f73496581e4feef655b4999a"
published: "2026-07-15T16:13:14.000Z"
score: 61
channel: "tech"
tags: ["EB-VAE","VAE","Time-to-Event Modeling","Cancer Genomics","Empirical Bayes","Pharmacometrics"]
---

# Multimodal Empirical Bayes Variational Autoencoders for Joint Longitudinal and Time-to-Event Modeling

## At a glance

This paper proposes a multimodal empirical Bayes variational autoencoder (EB-VAE) framework for joint modeling of longitudinal tumor measurements, dropout times, and genetic covariates. The framework uses latent individual effects regularized by a covariate-conditioned prior and combines neural dynamics with a semi-mechanistic decoder to jointly predict tumor growth trajectories and dropout risk. Experiments on melanoma and breast cancer show improved individual-level predictions and identification of biologically plausible genetic indicators such as BRAF and NRAS.

## Key points

- Proposes EB-VAE framework for joint longitudinal and time-to-event modeling with informative dropout.
- Semi-mechanistic decoder recovers treatment-effect parameters consistent with nonlinear mixed-effects models.
- Genetic conditioning improves individual-level predictions; stability selection identifies BRAF, NRAS, and other genes.
- Validated on melanoma and breast cancer datasets.

## Full text

Longitudinal tumor measurements, dropout information, and genetic covariates provide complementary information about treatment response, but integrating these data sources within a single population modeling framework remains challenging. We extend the empirical Bayes variational autoencoder (EB-VAE) framework to joint longitudinal and time-to-event modeling and evaluate it on tumor growth data. The framework represents inter-individual variability using latent individual effects regularized by a covariate-conditioned empirical Bayes prior, while a decoder maps these latent effects to tumor-volume trajectories. To account for informative dropout, the decoder was augmented with a hazard model, yielding joint predictions of tumor growth and time to dropout. We further compared fully neural and hybrid semi-mechanistic decoder formulations and incorporated genomic covariates through a genetics-conditioned prior adaptation. The hybrid decoder recovered treatment-effect parameters broadly consistent with previously reported nonlinear mixed-effects estimates, while achieving prior predictive performance comparable to the neural decoder. The joint model reproduced both tumor-volume distributions and dropout patterns in held-out individuals, and genetic conditioning improved individual-level prior predictions in both cutaneous melanoma and breast cancer experiments. Stability selection identified several biologically plausible genetic indicators, including alterations in BRAF, NRAS, NF1, and MDM2. These results demonstrate that EB-VAE provides a flexible probabilistic framework for combining neural dynamics, mechanistic structure, time-to-event modeling, and high-dimensional covariates in pharmacometric applications.

---

Verify against the original before citing; curated by Astral Feed: https://feed.astralor.com/items/itm_f73496581e4feef655b4999a