英语轻松读发新版了,欢迎下载、更新

Machine learning finds combined biological and psychosocial data improve chronic pain prediction

2025-05-28 10:50:03 英文原文

作者:by Ingrid Fadelli, Phys.org

Using machine learning to uncover biological markers and psychosocial factors that predict chronic pain conditions
Bar plots show mean ROC-AUC scores for models classifying pain diagnoses using biological (left) and psychosocial (right) modalities, with 95% CIs. Overlaid points represent individual validation folds. Bubble heat maps display AUC scores by modality subcategory; color indicates absolute AUC and size reflects z score relative to other diagnoses. Credit: Credit: Nature Human Behaviour (2025). DOI: 10.1038/s41562-025-02156-y

Chronic pain conditions, long-lasting medical conditions characterized by persistent pain in specific areas of the body, are often difficult to diagnose and treat. Understanding the biological markers (i.e., genes and brain patterns), psychological characteristics and social factors linked to a greater risk of developing these conditions could be highly advantageous, as it could help to devise more effective strategies to diagnose them.

Researchers at McGill University and other institutes recently carried out a study aimed at identifying biomarkers and psychosocial factors associated with the development of . Their findings, published in Nature Human Behavior, were obtained by analyzing data from a large biomedical database, namely the UK Biobank, using advanced techniques.

"Our study started as an effort to identify reliable brain-based biomarkers for chronic pain using data from the UK Biobank, the largest brain imaging cohort available," Matt Fillingim, first author of the paper, told Medical Xpress. "We quickly found that these biomarkers could not reliably distinguish chronic pain from pain-free individuals.

"However, when applied to specific pain conditions like fibromyalgia and , the biomarkers showed greater promise, prompting us to integrate additional psychosocial factors and diverse biological data (blood tests, bone imaging, genetics) to better understand chronic pain and its associated conditions."

As part of their study, Fillingim and his colleagues analyzed data collected from more than 523,000 people and stored in the UK Biobank. This data included brain imaging scans, genetic profiles, bone imaging scans, blood tests and detailed psychosocial information.

Using machine learning to uncover biological markers and psychosocial factors that predict chronic pain conditions
Kaplan–Meier curves show the 15-year cumulative incidence of diagnosis, grouped by blood (biomarker) and psychosocial risk (high–high, high–low, low–high, low–low). Credit: Credit: Nature Human Behaviour (2025). DOI: 10.1038/s41562-025-02156-y

"Using machine learning, we identified patterns that predicted various chronic pain conditions," explained Fillingim. "A crucial part of our approach involved stratifying individuals based on their biological and psychosocial risk profiles, which allowed us to understand how each contributes individually, and how they interact synergistically, to influence the development of chronic pain conditions."

Using machine learning–based methods, Fillingim and his colleagues were able to uncover biomarkers and psychosocial factors linked to 35 medical conditions associated with pain, including rheumatoid arthritis and gout, or self-reported chronic pain in specific body parts (e.g., back pain, knee pain, etc.).

The researchers found that models that looked at both and were better at predicting the development of chronic pain conditions than models that only focused on biomarkers or psychosocial factors.

"While biological markers effectively identified specific medical conditions associated with pain, psychosocial factors best predicted the subjective experience of pain," said Fillingim. "This strongly supports the prominent biopsychosocial model of pain and highlights the need for a holistic approach to pain diagnosis and management."

In the future, these results could inform the development of more reliable strategies to estimate the risk that specific people will develop chronic pain conditions or to accurately diagnose these conditions when people approach doctors with early symptoms.

"We've spent a great deal of time gathering additional data from pain studies around the world," added Fillingim. "Our next step is to test whether these findings replicate in individuals from different regions, cultures, and sociodemographic backgrounds. This will help us understand how universal these biomarkers and psychosocial factors are and whether they need to be adapted for specific populations."

More information: Matt Fillingim et al, Biological markers and psychosocial factors predict chronic pain conditions, Nature Human Behaviour (2025). DOI: 10.1038/s41562-025-02156-y

© 2025 Science X Network

Citation: Machine learning finds combined biological and psychosocial data improve chronic pain prediction (2025, May 28) retrieved 28 May 2025 from https://medicalxpress.com/news/2025-05-machine-combined-biological-psychosocial-chronic.html

This document is subject to copyright. Apart from any fair dealing for the purpose of private study or research, no part may be reproduced without the written permission. The content is provided for information purposes only.

关于《Machine learning finds combined biological and psychosocial data improve chronic pain prediction》的评论


暂无评论

发表评论

摘要

Researchers at McGill University and other institutions used machine learning on data from the UK Biobank to identify biomarkers and psychosocial factors associated with chronic pain conditions. Their findings, published in Nature Human Behavior, show that models incorporating both biological markers and psychological factors are better at predicting chronic pain than those using only one type of data. The study analyzed over 523,000 individuals' data including brain imaging scans, genetic profiles, blood tests, and psychosocial information. Results highlight the importance of a biopsychosocial approach in diagnosing and managing chronic pain. Future work aims to validate these findings across diverse populations globally.

相关新闻