Diagnostic accuracy and feasibility of a net-based application for diagnosing common psychiatric disorders

A net-based application, which enables users to diagnose and manage common psychiatric disorders independently, was developed. Diagnostic accuracy and reliability of the application were compared with the Mini International Neuropsychiatric Interview among 274 adult outpatients. The screening sub-module of the application had high sensitivity, specificity and negative predictive values for most disorders, but low positive predictive values for several disorders. The criteria-based diagnostic sub-module demonstrated moderate to substantial agreement (kappa>0.50) for all 4 broad categories and 10 of the 18 individual disorders included. Diagnostic agreement was, however, low (kappa<0.4) for most of the disorders comprising the broad category of 'neurotic and stress-related disorders'. Low sensitivity was also observed among these disorders, though specificity, and positive and negative predictive values were high for most disorders. Inter-rater reliability of the application's diagnoses was high. Symptom-severity and functional status scores correlated significantly with those on standard scales. Average time taken was 5min for screening and 20min for detailed diagnostic assessment. A majority of the patients, their relatives and interviewers were satisfied with the assessment. The results suggest that with further refinement the application could be suitable for use as a net-based diagnostic tool for psychiatric disorders.