Scholarlink

<p>Despite advances in pharmacotherapy, many patients with inflammatory arthritis (IA) experience residual pain even in remission. Digital health applications (DHAs), when combined with machine learning, may help identify predictors of pain relief in real-world care. To identify predictors of clinically meaningful pain relief in IA using DHA data. We retrospectively analyzed 914 adults with rheumatoid arthritis, psoriatic arthritis, or spondyloarthritis using the Mida Rheuma App. Patients were categorized according to sufficient pain relief, defined as ≥ 30% reduction in pain intensity (visual analogue scale, 0–100) at 12&#xa0;weeks. Candidate predictors for pain relief included baseline demographics, diagnosis, disease duration, and psychosocial and lifestyle factors: fatigue (Brief Fatigue Inventory, BFI), psychological distress (Patient Health Questionnaire, PHQ-4), sleep and diet quality, physical activity, social support, and app engagement. Multivariable logistic regression and Random Forest models were fitted; non-linear effects were explored using Shapley Additive Explanations. 25.4% of patients achieved ≥ 30% pain reduction. Higher baseline pain intensity was the strongest predictor of response (adjusted odds ratio [OR] 1.59). Fatigue reduced the likelihood of improvement (BFI OR = 0.71), with a SHAP-defined threshold of BFI &gt; 6. Better sleep quality was associated with an increased likelihood of response (OR = 1.23). In exploratory SHAP analyses, the decline in response probability was steepest above BFI of approximately 6. Model discrimination was modest (receiver operating characteristic area under the curve ≈0.61). In IA, pain relief in digital care is associated with baseline pain severity, fatigue, and sleep quality. These findings should be interpreted as hypothesis-generating and require external validation before clinical application. </p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Baseline pain, fatigue, and sleep quality predict 12-week pain improvement in inflammatory arthritis: retrospective real-world analysis of a digital health application cohort

  • 期刊论文

  • 出版模式：

    开放获取

  • 发表日期：

    2026年4月10日

查看全文
  • Dmytro Fedkov,
  • Danylo Yevstifeiev,
  • Oleg Iaremenko,
  • Daria Koliadenko,
  • Liubov Petelytska,
  • Christine Peine,
  • Felix Lang,
  • Abdullah Khalil,
  • Türker Kurt,
  • Stefan Vordenbäumen

摘要

Despite advances in pharmacotherapy, many patients with inflammatory arthritis (IA) experience residual pain even in remission. Digital health applications (DHAs), when combined with machine learning, may help identify predictors of pain relief in real-world care. To identify predictors of clinically meaningful pain relief in IA using DHA data. We retrospectively analyzed 914 adults with rheumatoid arthritis, psoriatic arthritis, or spondyloarthritis using the Mida Rheuma App. Patients were categorized according to sufficient pain relief, defined as ≥ 30% reduction in pain intensity (visual analogue scale, 0–100) at 12 weeks. Candidate predictors for pain relief included baseline demographics, diagnosis, disease duration, and psychosocial and lifestyle factors: fatigue (Brief Fatigue Inventory, BFI), psychological distress (Patient Health Questionnaire, PHQ-4), sleep and diet quality, physical activity, social support, and app engagement. Multivariable logistic regression and Random Forest models were fitted; non-linear effects were explored using Shapley Additive Explanations. 25.4% of patients achieved ≥ 30% pain reduction. Higher baseline pain intensity was the strongest predictor of response (adjusted odds ratio [OR] 1.59). Fatigue reduced the likelihood of improvement (BFI OR = 0.71), with a SHAP-defined threshold of BFI > 6. Better sleep quality was associated with an increased likelihood of response (OR = 1.23). In exploratory SHAP analyses, the decline in response probability was steepest above BFI of approximately 6. Model discrimination was modest (receiver operating characteristic area under the curve ≈0.61). In IA, pain relief in digital care is associated with baseline pain severity, fatigue, and sleep quality. These findings should be interpreted as hypothesis-generating and require external validation before clinical application.