Background: Hand dermatitis associated fingerprint changes is a significant problem and affects fingerprint verification processes. This study was done to develop a clinically useful prediction model for fingerprint verification in patients with hand dermatitis.
Methods: A case-control study involving 100 patients with hand dermatitis. All patients verified their thumbprints against their identity card. Registered fingerprints were randomized into a model derivation and model validation group. Predictive model was derived using multiple logistic regression. Validation was done using the goodness-of-fit test.
Results: The fingerprint verification prediction model consists of a major criterion (fingerprint dystrophy area of ≥ 25%) and two minor criteria (long horizontal lines and long vertical lines). The presence of the major criterion predicts it will almost always fail verification, while presence of both minor criteria and presence of one minor criterion predict high and low risk of fingerprint verification failure, respectively. When none of the criteria are met, the fingerprint almost always passes the verification. The area under the receiver operating characteristic curve was 0.937, and the goodness-of-fit test showed agreement between the observed and expected number (P = 0.26).
Conclusions: The derived fingerprint verification failure prediction model is validated and highly discriminatory in predicting risk of fingerprint verification in patients with hand dermatitis.
© 2014 The International Society of Dermatology.