FTIR spectroscopy-based differentiation of seborrheic dermatitis and scalp psoriasis using clinically normal fingernails.

OBJECTIVE

To establish a differential diagnostic method for scalp psoriasis (PsO) and seborrheic dermatitis (SD) using Fourier-transform infrared (FTIR) spectroscopy analysis of clinically normal fingernails.

METHODS

Clinically normal fingernail samples were collected from 60 subjects each in the healthy control (HC) group, scalp PsO group, and SD group as the training set, with an additional 30 subjects (9 scalp PsO, 11 SD, and 10 HC) collected as an independent validation set. Fingernail samples were analyzed using micro-FTIR spectroscopy. Feature selection was performed using interval partial least squares (iPLS) optimization, and a partial least squares discriminant analysis (PLS-DA) model was constructed. Model performance was evaluated through leave-one-out cross-validation.

RESULTS

The differential model for scalp PsO versus SD achieved a validation accuracy of 76.7% in the training set, with an area under the curve (AUC) of 0.889, sensitivity of 73.3%, and specificity of 80.0%. The independent validation set achieved an overall accuracy of 85.0%, AUC-ROC of 0.93, sensitivity of 90.9%, and specificity of 77.8%. The PsO vs HC model achieved an external validation accuracy of 94.7% (AUC = 0.99). The model demonstrated excellent robustness across different disease phenotypes and was unaffected by clinical confounders including pruritus severity, BMI, disease duration, and lifestyle factors.

CONCLUSIONS

FTIR spectroscopy analysis of clinically normal fingernails can effectively differentiate scalp PsO from SD, providing a simple, rapid, and non-invasive auxiliary diagnostic method for clinical practice.