Effect of diagnostic predictions combined with clinical information on avoiding perceptual errors of computed...

We hypothesized that even with appropriate clinical information, abnormal CT findings can still be missed if correct diagnostic predictions are not made.

Of 388 total students (97 5th-year medical students × 4), students who detected abnormalities without clinical information were eliminated. The remaining students (hereafter, subjects) obtained clinical information, made diagnostic predictions, and reevaluated images. The proportion of failures in detecting abnormalities was compared between the correct prediction group and the incorrect prediction group. In the correct prediction group, the relationship between failures of detection and the ranking of the correct diagnosis was also examined.

A total of 341 subjects were assessed. The proportion of subjects who failed to detect abnormalities in the correct prediction group (47.7 %, 93/195) was significantly lower (P < 0.001) than in the incorrect prediction group (85.6 %, 125/146). In the correct prediction group, the proportion of subjects who failed to detect abnormalities was significantly lower (P = 0.004) when the correct diagnosis was ranked first (38.5 %, 42/109) compared with lower rankings (59.3 %, 51/86).

Making appropriate diagnostic predictions and estimating the possibility of them based on clinical information is important to avoid missing abnormal CT findings.