The GMS-AGECAT system (Geriatric Mental State Schedule-Automated Geriatric Examination for Computer Assisted Taxonomy) has four stages. First, the patient is interviewed and rated on 541 symptoms. The scores are then condensed into 157 symptom components which are then grouped into clusters (organic; schizophrenia; depression; mania; obsessional, phobic, hypochondriacal, and anxiety neurosis), with each cluster assigned a level of confidence. Finally, the confidence levels are compared to determine a diagnosis. The authors state that a major problem of such systems is that verbal descriptions to users inadequately reflect the actual workings of the program. They have sought to overcome this by automatically generating descriptions from the same text that generates the program so that their descriptions are in step with the program and have the precision of an algorithmic language.

As noted by the authors, “a diagnostic system has many similarities with a scientific theory.” Here, then, the theory must be that algorithmic language and data manipulation are the proper way to comprehend an individual’s mental illness. I do not accept that 541 scores obtained in an interview with a patient are meaningful mental health professional-patient interactions. The “precision” of algorithmic language is surely spurious in the realm to which it is being applied and, even more important, its use skews the focus and concern of the user from that which is grounded in the intellectual, conceptual framework of his field to that amenable to computers.

I recommend that any potential user of this system first carefully read and consider the impassioned work of Joseph Weizenbaum [1].