AI-driven prediction of chronic pain after severe wounds: a pilot open-label study

Abstract

Background. Severe traumatic wounds, particularly in military settings, are frequently complicated by chronic, neuropathic, and phantom limb pain. Early prediction of pain trajectories remains a clinical challenge. Advances in artificial intelligence (AI) enable integration of wound imaging, microbiology, and pharmacological data into predictive models. We developed and tested an AI-driven platform for the prediction of chronic and neuropathic pain after severe wounds. Materials and methods. A prospective observational study of 311 Ukrainian military patients with severe limb injuries (including 139 amputees) was conducted in 2022–2025. Clinical, demographic, and microbiological data were collected alongside serial wound photographs. The platform processed wound images to assess healing dynamics, identify infection-related risk, and compute the probability of chronic, neuropathic, and phantom pain. Pain outcomes were assessed at 3 and 6 months using the Numeric Rating Scale (NRS) and the DN4 questionnaire. Results. At 6 months, chronic pain was present in 42 % of patients, neuropathic pain in 29 %, and phantom pain in 24 % of amputees. Independent predictors of neuropathic pain included wound infection (odds ratio (OR) 2.1, 95% confidence interval (CI) 1.4–3.2), delayed wound healing (> 8 weeks) (OR 2.7, 95% CI 1.8–4.0), high baseline pain intensity (NRS ≥ 7) (OR 1.9, 95% CI 1.2–3.0), and exposure to neurotoxic antibiotics (OR 1.8, 95% CI 1.1–2.9). The platform achieved a sensitivity of 78 %, specificity 74 %, and AUC of 0.81. Conclusions. Infection, delayed healing, and neurotoxic drug exposure are major predictors of chronic and neuropathic pain after severe wounds. The AI platform provides accurate, clinically relevant risk prediction and may support personalized pain prevention and rehabilitation in military trauma care.