iLoRA: Bayesian Low-Rank Adaptation with Latent Interaction Graphs for Microbiome Diagnosis

2026-05-28

AI Summary

Existing low-rank adaptation (LoRA) methods struggle to model the latent microbial interaction patterns underlying scientifically meaningful labels—such as disease states—and lack dynamic responsiveness to input conditions. This work proposes iLoRA, a novel framework that integrates Bayesian inference with graph conditioning into LoRA for the first time. iLoRA generates input-conditioned low-rank updates and jointly learns both the prediction task and the latent interaction graph structure in an end-to-end manner, eliminating the need for post-hoc analysis while providing calibrated uncertainty estimates. Evaluated on multi-cohort inflammatory bowel disease diagnosis and interactive question-answering tasks, iLoRA significantly outperforms standard LoRA and Bayesian baselines, achieving higher diagnostic accuracy and reliably recovering expert-annotated microbial interactions.

Abstract

Parameter-efficient adaptation has made LLMs practical for domain prediction, but standard LoRA still relies on a static low-rank update and does not expose the latent interactions that often drive scientific labels. We introduce iLoRA. To our knowledge, it is the first Bayesian graph-conditioned LoRA framework. It infers a latent interaction graph from the input and uses it to generate input-conditioned LoRA updates. As a result, iLoRA learns prediction and latent interaction structure jointly, rather than training a predictor and applying interaction analysis only post hoc. We instantiate this idea for microbiome diagnosis, where disease state can depend on both species-level abundance and microbe-microbe cross-talk, and evaluate it in two complementary settings: interactive QA with human-annotated graphs, which tests latent structure recovery, and multi-cohort IBD diagnosis, which tests biomedical utility. Across both settings, iLoRA improves over strong LoRA and Bayesian adaptation baselines, recovers graphs aligned with human annotations and cohort-level microbiome associations, and provides calibrated uncertainty with moderate graph-branch overhead.

Problem

• microbiome diagnosis
• latent interaction
• low-rank adaptation
• Bayesian adaptation
• parameter-efficient tuning

Innovation

• Bayesian LoRA
• latent interaction graph
• parameter-efficient adaptation
• microbiome diagnosis
• graph-conditioned adaptation