A comparison of clinical and automated interpretation of the halstead-reitan battery

Abstract

The diagnostic accuracy of two clinicians was compared with that of the Russell, Neuringer, and Goldstein (1970) quasi-actuarial system for interpreting the Halstead-Reitan Battery (HRB). The clinicians' independent ratings of HRB data were highly reliable, and were more accurate than the automated system in predicting the presence and laterality of structural cerebral lesions. Comparisons of the subgroups correctly and incorrectly classified by the clinicians suggest some intrinsic limitations of the test data in predicting lesion presence, and especially laterality. Classification accuracy was related to etiology of lesions, and also to extent and location of focal lesions. Neither the clinicians nor the automated system exceeded base-rate predictions regarding chronicity, but this may have been due to sampling bias in the acute group. Although the clinicians' superiority in two of the three prediction tasks may be due to their greater flexibility in weighting complex and unique patterns of test data, it remains to be seen whether this potential advantage will hold up against more sophisticated actuarial methods.