A clinical decision support system (CDSS) was found to have diagnostic accuracy among patients with primary and secondary headache, according to an observational study published in The Journal of Headache and Pain.
Researchers identified 653 patients (mean age 38.36 [SD, 14.21] years, 406 women) from 16 hospitals in China for inclusion in this study. The CDSS 2.0 generated an automated diagnosis based on demographic information collected during a human to computer conversation, whereby patients responded to a questionnaire that included over 60 questions using their personal mobile devices. The CDSS 2.0 also issued a warning to patients with suspicious secondary headache.
Each patient was classified by the CDSS 2.0 first, then by a clinician who exclusively reviewed patient responses in the CDSS 2.0 (expert 1), and also by a clinician who reviewed outpatient medical records in addition to the CDSS 2.0 responses (expert 2).
The average CDSS 2.0 completion time was 10.95 minutes, and 97.5% of patients completed the questionnaire in under 30 minutes.
Overall, specialists recommended 18.68% of patients for a secondary headache screening.
Among the remaining 531 patients included in the diagnostic accuracy analysis, the CDSS 2.0 correctly recognized 80.00% to 100% of 12 out of 13 categories of headache and migraine, when compared with expert 1; the exception was neuralgia, which the CDSS 2.0 only identified in 57.78% patients. In addition, the CDSS 2.0 had a recognition rate between 75.29% and 96.15%, compared with expert 2.
Compared with expert 1, the CDSS 2.0 had the best diagnostic performance for infrequent episodic tension-type headache (TTH; k, 1.000; sensitivity, 1.0000; specificity, 1.0000), new daily persistent headache (NDPH; k, 1.000; sensitivity, 1.0000; specificity, 1.0000), and medication overuse headache (MOH; k, 0.981; sensitivity, 0.9655; specificity, 1.000).
Further, the CDSS 2.0 had the best diagnostic performance for chronic TTH (k, 0.933; sensitivity, 0.9565; specificity, 0.9961), MOH (k, 0.913; sensitivity, 0.8485; specificity, 1.0000), and chronic migraine (k, 0.854; sensitivity, 0.9000; specificity, 0.9962), compared with expert 2.
Results of a patient satisfaction survey revealed that 54.58% were very satisfied with operability of the CDSS 2.0; 52.58% were very satisfied with fluency; 51.29% with intelligibility; 48.71% with efficiency; 46.71% with feasibility; and 46.13% with comprehensiveness.
Study limitations included that no headache classifications were made through direct consultation with the patient, and that there was lack of treatment and follow up.
The researchers concluded, “The CDSS 2.0 demonstrated a high degree of accuracy in recognizing migraines, TTH, [chronic headache], NDPH, and MOH but not neuralgia. Given the high diagnostic accuracy for most of the primary headache types and a satisfactory ability to alert for secondary headaches, this system is likely to improve diagnostic accuracy for headache disorders and thereby reduce the burden of headache-related conditions in China.”
Han X, Wan D, Zhang S, et al. Verification of a clinical decision support system for the diagnosis of headache disorders based on patient–computer interactions: a multi‑center study. J Headache Pain. Published online May 23, 2023. doi:10.1186/s10194-023-01586-1