Luo, Xuexing, Li, Yiyuan, Xu, Jing et al. · Journal of medical Internet research · 2025 · DOI
Quick Summary
Researchers reviewed 14 studies about using artificial intelligence (AI) to improve medical questionnaires—the forms doctors use to diagnose conditions. They found that AI can help doctors distinguish ME/CFS from long COVID with over 92% accuracy, develop better questionnaires, and predict disease risk. However, most research is still early-stage, and more testing is needed before AI tools are ready for everyday clinical use.
Why It Matters
For ME/CFS patients and researchers, this review highlights AI's potential to improve diagnostic accuracy and differentiate ME/CFS from similar conditions like long COVID—a critical challenge since misdiagnosis delays appropriate treatment. The finding that AI can assess questionnaires with high accuracy suggests promise for developing better diagnostic tools specific to ME/CFS, though the field remains largely exploratory.
Observed Findings
AI distinguished ME/CFS from long COVID with 92.18% accuracy in assessment tasks.
Only 3 of 14 studies (21%) had entered clinical validation phase; 11 (79%) remained in exploratory phase.
10 of 14 studies (71%) were rated moderate methodological quality with significant limitations.
24 distinct AI technologies were identified, ranging from traditional algorithms (random forest, support vector machine) to deep learning models (convolutional neural networks, BERT, ChatGPT).
Natural language processing with generative models successfully developed culturally competent assessment scales.
Inferred Conclusions
Integrated AI application in medical questionnaires has significant potential to improve diagnostic efficiency and accelerate scale development.
Current AI research in this domain remains predominantly exploratory, with substantial methodological gaps limiting clinical translation.
Future advancement requires prioritizing model interpretability, standardized validation, ethical governance, and system compatibility to address data privacy and transparency concerns.
Remaining Questions
How do AI-enhanced questionnaires perform in prospective clinical validation studies with ME/CFS patients across diverse geographic and socioeconomic populations?
What This Study Does Not Prove
This review does not prove that AI questionnaire tools are ready for clinical use—only 21% of identified studies have reached clinical validation. The review also does not establish how well these AI approaches would work specifically for ME/CFS diagnosis in real-world clinical settings, nor does it demonstrate that improved questionnaire assessment directly translates to better patient outcomes.
About the PEM badge: “PEM required” means post-exertional malaise was an explicit required diagnostic criterion for participant inclusion in this study — not that PEM was studied, observed, or discussed. Studies using criteria that do not require PEM (e.g. Fukuda, Oxford) are tagged “PEM not required”. How the atlas works →