Diagnostic sensitivity of NeuroAdvise, an innovative medical application in diagnosis of neurological disorders

Background and aims: In recent years, artificial intelligence-based computerized systems arebeing widely used by doctors in their clinical practice.Doctors need to be ensured regarding sensitivity and accuracy of the diagnoses and recommendationsof a computerized system to trust and use it.In this study, we have evaluated the sensitivity of an innovative computer-assisted diagnosticsystem developed for diagnosis and evaluation of neurological disorders.Method: We reviewed documents of two groups of patients including ۱۰۰ patients with a definitediagnosis of ischemic stroke as a common disease and ۵۳ patients with a diagnosis of multiplesclerosis as a relatively uncommon disorder who were admitted to Imam Reza Hospital in ۲۰۱۸.The discharge note , CT scan and MRI examination report related to each patient were extractedand their clinical data entered to our innovative clinical decision support system, called “NeuroAdvise”that is available as an Android application.. The inclusion criteria for this study werepatients hospitalized in ۲۰۱۸ with a definite diagnosis of ischemic stroke and multiple sclerosis,and the exclusion criteria were incomplete clinical documents. NeuroAdvise is an intelligentsystem with different approaches that allows the user to diagnose the disease according to theclinical information entered into it. In this study we evaluated “anatomic approach” and “clinicalapproach”. The sensitivity of this system was evaluated by defining the percentage of patients thatwere correctly reached the definite diagnosis by using the system.Results: NeuroAdvise system diagnosed the ischemic stroke as the first diagnosis in ۹۵% ofcases and one out of top three diagnoses in ۹۹% of cases. Accordingly, multiple sclerosis wasthe first diagnosis in ۴۴.۸% of cases, and was within top three diagnoses in ۷۵.۴۵% of cases. Byconsidering the complete list of differential diagnosis provided by the system for each patient, thesensitivity was ۱۰۰% for both groups of ischemic stroke and multiple sclerosis patients.Conclusion: NeuroAdvise expert system, is more sensitive in diagnosis of common disorders andmore specific in diagnosis of more rare disorders. NeuroAdvise, as an smart assistant, can providea practical guide to doctors in the field of examination, differential diagnosis and step-by-step investigationand can help in correct and timely diagnosis of neurological disorders. NeuroAdvisecan reduce medical errors, improve neurology patients care and reduce unnecessary costs.Such intelligent systems, cannot currently replace the physicians, but can help them in makingbetter clinical decisions.

کلیدواژه ها:

Artificial Intelligence ، NeuroAdvise ، CDSS (Clinical Decision Support System) ، Medicine