Reproduce “Genetics-Driven Personalized Disease Progression Model” on OpenResearch
Turn your favorite coding agents into research agents that review literature, develop hypotheses, and run experiments.
Universal app for Apple silicon and Intel. Requires macOS 11 or later.
About this paper
Haoyu Yang, Sanjoy Dey, Pablo Meyer
Researchers from the University of Minnesota and IBM Research developed PerDPM, a genetics-driven personalized disease progression model that integrates a variational autoencoder for genetic inference with an RNN-based state-space model. The model jointly learns distinct genetic subgroups and their specific disease progression trajectories, achieving superior performance on both synthetic data and real-world Chronic Kidney Disease progression modeling from UK Biobank data compared to baseline methods.
Key insights
- Genetic information fundamentally drives distinct disease progression dynamics, requiring explicit modeling beyond simple static covariate inclusion.
- Jointly learning patient genetic subgroups and their associated progression patterns yields more accurate and interpretable personalized disease trajectories.
- A soft clustering approach via VAE, combined with an attention mechanism over cluster-specific transition functions, effectively personalizes state-space models based on genetic makeup.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.