Glyph: An Interpretability Engine for AI-Driven Scientific Discovery in Clinical Data

Hugging Face Daily Papers Papers

Summary

Glyph is an automated multistage framework that uses foundation models and sparse autoencoders to extract interpretable clinical concepts from multimodal records and CT data, identifying 397 predictive concepts with thousands of reproducible associations across patient cohorts.

Automated scientific discovery has the potential to advance understanding of disease mechanisms and accelerate identification of early disease markers. Manual data annotation is infeasible at population scale, motivating methods that automatically extract pre-defined phenotypes in multimodal and unstructured clinical data. Such pre-defined feature sets determine and limit which associations these approaches can discover. This limitation extends to agentic AI systems that use these tools to query population-scale data. In contrast, foundation models (FMs) trained on large volumes of clinical data can perform tasks beyond their training objectives. This suggests their latent embeddings may represent novel or unexplored features that could yield insights into disease mechanisms, yet such latent features remain difficult to interpret. To address this challenge, we introduce Glyph, an automated multistage framework that extracts features from multimodal and unstructured clinical data using FMs and sparse autoencoders, interprets them as coded clinical concepts, and evaluates their value for clinical scientific discovery. We applied Glyph to abdominal CT volumes, CT radiology reports and laboratory measurements from 41,628 patients across different institutions. We identified 397 predictive concepts supported by 4,463 Bonferroni-significant associations. Moreover, a majority of discovered associations were automatically reproducible in held-out cohorts across different sites. Glyph concepts provided complementary associations to literature-derived markers, leading to gains in predictive power. We envision that interpretability can extend beyond a mechanistic understanding of neural networks to connecting implicitly learned features with clinical knowledge and establishing their significance for patient outcomes. We release Glyph at https://github.com/StanfordMIMI/Glyph.
Original Article
View Cached Full Text

Cached at: 11/19/26, 03:41 AM

Paper page - Glyph: An Interpretability Engine for AI-Driven Scientific Discovery in Clinical Data

Source: https://huggingface.co/papers/2611.08922 Authors:

,

,

,

,

,

,

,

,

,

,

,

,

,

,

,

,

,

,

,

Abstract

Glyph uses foundation models and sparse autoencoders to extract interpretable clinical concepts from multimodal records and CT data, revealing reproducible predictive associations across patient cohorts.

Automated scientific discovery has the potential to advance understanding of disease mechanisms and accelerate identification of early disease markers. Manual data annotation is infeasible at population scale, motivating methods that automatically extract pre-defined phenotypes in multimodal and unstructured clinical data. Such pre-defined feature sets determine and limit which associations these approaches can discover. This limitation extends toagentic AI systemsthat use these tools to query population-scale data. In contrast,foundation models(FMs) trained on large volumes of clinical data can perform tasks beyond their training objectives. This suggests theirlatent embeddingsmay represent novel or unexplored features that could yield insights into disease mechanisms, yet such latent features remain difficult to interpret. To address this challenge, we introduce Glyph, an automated multistage framework that extracts features from multimodal and unstructured clinical data using FMs andsparse autoencoders, interprets them as coded clinical concepts, and evaluates their value for clinical scientific discovery. We applied Glyph to abdominal CT volumes, CT radiology reports and laboratory measurements from 41,628 patients across different institutions. We identified 397 predictive concepts supported by 4,463 Bonferroni-significant associations. Moreover, a majority of discovered associations were automatically reproducible in held-out cohorts across different sites. Glyph concepts provided complementary associations to literature-derived markers, leading to gains in predictive power. We envision thatinterpretabilitycan extend beyond a mechanistic understanding ofneural networksto connecting implicitly learned features with clinical knowledge and establishing their significance for patient outcomes. We release Glyph at https://github.com/StanfordMIMI/Glyph.

View arXiv pageView PDFProject pageGitHub2Add to collection

Get this paper in your agent:

hf papers read 2611\.08922

Don’t have the latest CLI?curl \-LsSf https://hf\.co/cli/install\.sh \| bash

Models citing this paper1

#### RobbieHolland/Glyph-research Feature Extraction• Updated9 days ago

Datasets citing this paper1

#### RobbieHolland/Glyph-research Viewer• Updated9 days ago • 882k • 116

Spaces citing this paper0

No Space linking this paper

Cite arxiv.org/abs/2611.08922 in a Space README.md to link it from this page.

Collections including this paper0

No Collection including this paper

Add this paper to acollectionto link it from this page.

Similar Articles

Generalized Biomedicine Discovery

arXiv cs.LG

The paper introduces Generalized Biomedicine Discovery (GBD), a unified benchmark for open-world medical imaging that handles long-tailed rare diseases, subtle lesions, and hierarchical taxonomies, and proposes SCAN, a cognition-inspired method using predictive suppression, surprise-evoked saliency, and complementary accommodation to improve novel concept discovery while preserving known clinical knowledge.