Profiling of extracellular vesicles from primary hepatocytes, organoids, and mash patients identifies cell injury-specific signatures

2026-06-03 · Source: Machine learning : nature.com subject feeds · Field: Health & Wellbeing — Health & Medical Research, Medical Devices & Health Technology, Artificial Intelligence & Machine Learning · Depth: Expert, extended

Summary

This study identifies non-invasive extracellular vesicle (EV) protein signatures for Metabolic Dysfunction-Associated Steatohepatitis (MASH), a severe form of Metabolic Dysfunction-Associated Steatotic Liver Disease (MASLD). Researchers assayed 6596 proteins from 38 individuals with histologically confirmed MASLD, utilizing patient serum, primary human hepatocytes (PHH), and human liver organoids (HLO). Key proteins, including SLC27A5, HP, and CXCL7, were elevated in patient samples, while PHH and HLO models exhibited upregulation of ASGPR1, HP, and CXCL7 under MASH conditions. A deep learning model, trained with 19 essential protein markers, achieved an AUROC of 0.97 and 97.5% accuracy in classifying MASH from healthy controls, validated on an independent bariatric surgery cohort. This integrative approach advances MASH biomarker discovery, linking localized liver dysfunction with systemic disease mechanisms for non-invasive diagnostics and personalized treatments.

Key takeaway

For AI Scientists and Clinical Researchers developing non-invasive diagnostics for liver diseases, this study demonstrates a robust methodology. You should explore integrating multi-model EV proteomic data with deep learning to identify highly accurate disease signatures. This approach offers a path to overcome limitations of traditional biopsies, enabling earlier detection and personalized treatment strategies for conditions like MASH, potentially improving patient outcomes and monitoring therapeutic response.

Key insights

Extracellular vesicle protein signatures, identified across multi-models and validated by machine learning, offer non-invasive MASH diagnostics.

Principles

• EV protein profiles reflect cell health and disease stages.
• Multi-model integration enhances biomarker robustness.
• Machine learning refines complex proteomic data.

Method

The study used aptamer-based technology to assay 6596 proteins from patient serum, PHH, and HLO EVs. Machine learning, specifically a neural network with Ridge regression for feature selection, classified MASH with high accuracy.

In practice

• Investigate EV protein panels for early MASH detection.
• Apply deep learning to complex proteomic datasets.
• Consider HP, SLC27A5, CXCL7 as MASH indicators.

Topics

• Metabolic Dysfunction-Associated Steatohepatitis
• Extracellular Vesicles
• Proteomics
• Biomarker Discovery
• Deep Learning
• Non-invasive Diagnostics

Code references

• aleksanarepo/Protein-Based-Disease-Classification-Model

Best for: Research Scientist, AI Scientist, Data Scientist

Related on AIssential

• Non-invasive profiling of the tumour microenvironment with spatial ecotypes — A new machine learning framework enables non-invasive profiling of tumor microenvironment spatial ecotypes from liquid biopsies, predicting immunotherapy response.
• Accelerating discovery of liver disease mechanisms — AI-driven literature analysis can uncover novel biological hypotheses and candidate therapies for complex diseases.
• Deep learning-based in silico labeling for analyzing morphological features of MSCs to predict immunomodulatory capacity — A dual AI model non-invasively predicts MSC immunomodulatory capacity from morphology, enhancing cell therapy quality control.
• Machine Learning-Based Pre-Test Risk Stratification for PCR-Confirmed Chlamydia Using Patient-Reported Data and Urine Biomarkers — Urine biomarkers offer a reliable, consistent signal for Chlamydia pre-test risk stratification, complementing patient-reported data.
• Bringing GigaTIME to Microsoft Foundry: Unlocking Tumor Microenvironment Insights with Multimodal AI — GigaTIME translates routine pathology slides into detailed protein activation maps for tumor microenvironment analysis.
• 3 Questions: Building predictive models to characterize tumor progression — AI and ML tools can decode tumor evolution, especially ecDNA amplifications, to predict and counter treatment resistance.

Open in AIssential →

Editorial summary, takeaway, and curation by AIssential. Original article published by Machine learning : nature.com subject feeds.