[Skip to Content]
[Skip to Content Landing]
August 1974

Constraints on the Validity of Computer Diagnosis

Author Affiliations

From Biometrics Research, New York State Department of Mental Hygiene, New York State Psychiatric Institute; and the departments of psychiatry (Drs. Spitzer and Endicott) and biostatistics (Dr. Fleiss), Columbia University; and the Department of Psychology (Dr. Cohen), New York University.

Arch Gen Psychiatry. 1974;31(2):197-203. doi:10.1001/archpsyc.1974.01760140049008

Issues in the constraints on the validity of computerized psychiatric diagnosis are illustrated by the analysis of diagnoses produced by treating clinicians, expert diagnosticians, and the DIAGNO III computer diagnostic system. The results indicate modest agreement between the computerized diagnoses and both clinicians and experts, and not much better agreement between the experts and between the treating clinicians. The main constraint on the validity of computerized diagnoses is not in any inherent limitation on computer processing but rather in the limitations of the current diagnostic system itself. Improvements in computer diagnosis await improvements in the diagnostic system, along the lines of simplification, explicit criteria, and limitation of the categories to those conditions for which validity evidence exists.