Modular algorithm for tardive dyskinesia diagnosis (MALTD): a demonstration of a methodological concept for diagnostic decision making.

ABSTRACT

A set of algorithms has been written which assign the Research Diagnoses for Tardive Dyskinesia (RD-TD) to patients evaluated for this neurological syndrome that occurs as a side effect of neuroleptic drug treatment. The approach used in constructing these algorithms does not employ decision tree and flowcharting mechanisms often used in making automated diagnoses. Rather, a modular structure is used where a value is assigned to each independent component relevant to tardive dyskinesia, e.g. presence or absence of a minimal amount of abnormal involuntary movements, length of time of neuroleptic drug-taking. A particular combination of values of these components determines the diagnosis assigned. Advantages of such a modular system structure are that it permits new components to be added, existing components to be deleted, the criterion levels for existing components to be modified, and/or the definition of any of the diagnostic labels to be altered without affecting other components. Although this algorithm was developed specifically for research in tardive dyskinesia, the authors hope that the concepts of the modular structure will apply to the study of other longitudinal data.