Huge thanks to our incredible team and collaborators.

Read the full paper here: nature.com/articles/s41...

Grateful to Nature Human Behaviour for publishing this work.

#ChronicPain #PrecisionMedicine #Biomarkers #Psychosocial #MachineLearning

Adding psychosocial context dramatically improved prediction accuracy across all pain phenotypes.
🧬 + 📋 = 🔍

This synergy paints a richer picture of pain vulnerability and brings us closer to personalized pain care.

We created biomarker and psychosocial risk scores and grouped participants into quintiles.

Those high on both risks had over 2× higher incidence of painful conditions over 15 years, while those high on just one showed little to no added risk.

Biomarkers alone accurately predicted many painful medical conditions, often outperforming psychosocial models.

But for self-reported pain, biology wasn’t enough, psychosocial models performed significantly better.

We then compared the biological models to psychosocial models spanning mental health, physical well-being, and sociodemographic factors.

We applied machine learning to four biological data types:🩸blood assays, 🦴bone scans, 🧠brain imaging, and 🧬genetics, to develop biomarkers for conditions like arthritis or migraine, as well as self-reported bodily pain.

We asked: Can combining biological and psychosocial information improve prediction of chronic pain conditions?

Spoiler: Yes, significantly.