Diagnosis of chronic fatigue syndrome using beat-to-beat autonomic measurements.
Kujawski, Sławomir, Tabisz, Hanna, Morten, Karl J et al. · Journal of translational medicine · 2025 · DOI
Quick Summary
Researchers used artificial intelligence to analyze how the heart and nervous system behave differently in ME/CFS patients compared to healthy people. By measuring heart rate changes beat-by-beat, they found a pattern that could identify ME/CFS patients with 89% accuracy. The key differences included a less responsive vagal nerve (which normally calms the body), a more active stress response in blood vessels, and less efficient heart pumping.
Why It Matters
This study provides the first objective, AI-based diagnostic tool with high accuracy for ME/CFS using non-invasive autonomic measurements, potentially enabling faster diagnosis and reducing diagnostic delays that patients typically experience. Understanding the specific autonomic dysfunction patterns may open new avenues for targeted treatment strategies. The findings validate long-standing clinical observations that autonomic dysfunction is central to ME/CFS pathology.
Observed Findings
AI classifier achieved 89% subject-level accuracy in distinguishing CFS patients from healthy controls
CFS patients exhibited lower stroke volume and stroke volume index values (impaired cardiac hemodynamics)
Model achieved perfect ROC AUC of 1.00, though this requires validation in external cohorts
Inferred Conclusions
Beat-to-beat autonomic nervous system measurements can objectively identify ME/CFS patients with high diagnostic accuracy
ME/CFS is characterized by a specific pattern of autonomic dysfunction: reduced parasympathetic (vagal) tone combined with elevated sympathetic vascular tone and impaired cardiac pumping efficiency
Automated AI-driven analysis of high-frequency ANS data may enhance objective assessment and diagnosis of ME/CFS in clinical settings
Autonomic dysfunction assessment could serve as a biomarker for ME/CFS diagnosis and potentially for disease monitoring
Remaining Questions
Does the validated model maintain high accuracy when tested on independent external cohorts from different centers and populations?
What This Study Does Not Prove
This study does not establish causation—it demonstrates that these autonomic patterns associate with ME/CFS but does not prove they cause the disease. The perfect AUC warrants caution regarding overfitting and requires external validation in independent cohorts before clinical implementation. The cross-sectional design cannot determine whether these autonomic changes precede disease onset or develop as a consequence of ME/CFS.
Are these autonomic patterns specific to ME/CFS or do they overlap with other conditions involving autonomic dysfunction (e.g., POTS, Long COVID, dysautonomia)?
Do these beat-to-beat autonomic measurements correlate with symptom severity, functional capacity, or treatment response in ME/CFS patients?
Can serial measurements track disease progression or predict treatment outcomes, and do autonomic patterns normalize with recovery?