Multifactorial Machine Learning Algorithm Integration of Pain Mechanisms Can Predict the Efficacy of 3‐Week <scp>NSAID</scp> Plus Paracetamol in Patients With Painful Knee Osteoarthritis
European Journal of Pain, 2025
Giordano R., Arendt‐Nielsen L., Hertel E., Olesen A., Petersen K.
| Disease area | Application area | Sample type | Products |
|---|---|---|---|
Neurology | Patient Stratification | Plasma |  Olink Target 96 |
Abstract
Background
Studies demonstrate that pain sensitization, epigenetic mechanisms, inflammation, and psychological factors might be predictive of treatment outcomes. Anti‐inflammatory therapy is recommended, but efficacy varies among patients. This study aimed to utilise machine learning to predict the analgesic responses of 3‐week NSAID plus paracetamol therapy using pre‐treatment assessments of pain sensitivity, inflammation, microRNA, and psychological factors.
Methods
Patients (n = 101) underwent 3‐week combined NSAID plus paracetamol therapy. Pain sensitivity using cuff algometry, Hospital Anxiety and Depression Scale, Pain Catastrophizing Scale, EQ‐5D‐3L scale, and blood samples were collected before therapy. Pain relief was assessed by the Knee Injury and Osteoarthritis Outcome Score pain subscale, before and after therapy. Inflammatory biomarkers were analysed using Olink, and microRNA using Next‐Generation RNA Sequencing. Data Integration Analysis for Biomarker discovery using Latent cOmponents (DIABLO) was utilised to integrate the pre‐treatment data and explain the analgesic effect.
Results
DIABLO model identified 30 significant variables across the 4 domains. After cross‐validation, model performance showed an area under the precision‐recall curve of 85%, sensitivity of 83%, specificity of 87%, and balanced accuracy of 85%.
Conclusions
This study utilises a machine learning algorithm, based on pain sensitization, epigenetics, inflammatory response, and psychological factors, to predict analgesic response in osteoarthritis patients. The study demonstrates that incorporating multiple factors into a model enhances its performance, enabling the identification of patients who will benefit from therapy, advancing personalised pain management.
Significance Statement
In this study, a machine learning algorithm, based on pain sensitization, epigenetic mechanisms, inflammatory response, and psychological factors, predicts analgesic response in osteoarthritis patients with 84% accuracy.
Trial Registration
ClinicalTrials.gov identifier: NCT02967744