A MACHINE-learning classifier designed to assist healthcare professionals in accurately diagnosing the causes of transient loss of consciousness (TLOC) has been introduced in a new study. TLOC, commonly presenting as sudden fainting, can stem from various conditions, including syncope, epilepsy, or functional/dissociative seizures, making prompt and precise diagnosis challenging.
The research team conducted a prospective study involving 178 patients who presented with TLOC at emergency departments, acute medical units, and specialized clinics. Participants completed an online questionnaire detailing their experiences, either at home or during their initial medical assessment. After a six-month follow-up, two expert raters determined the underlying cause of each patient’s TLOC. Utilizing these questionnaire responses, the researchers developed a random forest classifier—a type of machine-learning model—to predict diagnoses and validated its performance against traditional diagnostic methods.
The classifier, based on nine specific questionnaire items, correctly identified 80.8% of cases (63 out of 78 diagnoses; 95% CI 70.0–88.5). In comparison, initial assessing clinicians achieved a 70.5% accuracy rate. Notably, the classifier demonstrated a sensitivity of 96% (87.0%–99.4%) in correctly identifying syncope cases. However, its specificity was 40% (20%–63.6%) for accurately classifying epilepsy or functional/dissociative seizures as non-syncope.
Despite these promising results, the study acknowledges that the classifier’s current accuracy is insufficient for routine clinical application. The researchers suggest that incorporating information from witnesses could enhance the system’s diagnostic performance, potentially leading to more reliable differentiation between the various causes of TLOC.
This advancement underscores the potential of integrating artificial intelligence into clinical decision-making processes. By leveraging machine-learning algorithms, healthcare professionals may soon have access to tools that augment diagnostic accuracy, particularly in complex cases like TLOC. Continued research and development are essential to refine these technologies, ensuring they meet the rigorous standards required for everyday medical practice.
Reference: Wardrope A et al. Validation of a Machine-Learning Clinical Decision Aid for the Differential Diagnosis of Transient Loss of Consciousness. Neurol Clin Pract. 2025;25(2).
Anaya Malik | AMJ
