Accuracy and safety of an artificial intelligent system for nonacute headache diagnosis

DORR, FRANCISCO; GOICOCHEA, TERESA; GRIMALDI, FRANCISCO; ACOSTA, JULIAN; ALESSANDRO, LUCAS; VARELA, FRANCISCO; FAREZ, MAURICIO; FERNÁNDEZ SLEZAK, DIEGO

Congreso; American Congress of Neurology; 2018

Objective: Evaluate accuracy and safety of an artificial intelligent (AI) system for nonacute headache diagnosis.Background: Headache is the main cause of neurologic consultation, entailing high cost in healthcare systems and a great impact in quality of life of patients suffering from it. Moreover, the access to qualified specialists and appropriate treatment is not ensured, especially in areas with low number of neurologist per capita. We hypothesize that and AI-system could assist in the diagnosis of headaches with a precision and safety comparable to a specialist.Design/Methods: We reviewed a database of 580 clinical records of patients with headache as chief complaint. Clinical records were processed with Latent Semantic Analysis (LSA) and a Support Vector Machine (SVM) model was trained. The definite diagnosis was the one given by the specialist at the consultation. We compared the SVM model performance at classifying the headache as primary versus secondary with two general neurologist.Finally, we used an interactive headache questionnaire filled by patients previous to the consultation and classified the headache with an automatic ICHD criteria system supplemented with a machine-learning model, comparing that diagnosis to the one given by neurologists. All the development and analysis was done using Python.Results: The SVM model trained after ?reading? clinical records with LSA had a better performance in the diagnosis of secondary headache (sensitivity=90.2%; specificity=93%) in comparison with other neurologists (sensitivity=82%; specificity=85%). A correct headache diagnosis was achieved in 89?94% of the cases when ICHD criteria was combined with several machine-learning models.Conclusions: AI has a great potential for its application in headache diagnosis. Advancements in this field would both improve the accessibility to quality healthcare and optimize the time spent by health professionals.