Cumulative Sum Analysis-Integrated E-Learning for Differentiation Between Basal Cell Carcinoma and Non-Basal Cell...

The clinical implementation of optical coherence tomography (OCT) for diagnosing clinically equivocal lesions suspicious for basal cell carcinoma (BCC) is limited by an OCT assessor shortage. Cumulative sum (CUSUM) analysis enables monitoring of diagnostic performance during training. The objective was to evaluate whether CUSUM-integrated e-learning is suitable for training healthcare professionals in achieving and maintaining an acceptable error rate for differentiating BCC from non-BCC lesions on OCT. Furthermore, we explored the diagnostic accuracy of high-confidence BCC diagnoses by the newly trained assessors.

A CUSUM-integrated e-learning was developed. Trainee performance was monitored by CUSUM analysis. The number of OCT scans required to achieve and maintain a predefined acceptable error rate (percentage of correct diagnoses) was evaluated. Successfully trained assessors were asked to discriminate BCC from non-BCC on 100 OCT scans for a subsequent diagnostic accuracy study. Histopathology served as the reference standard.

Seventeen trainees successfully completed the training. Adequate performance was achieved and maintained after assessing a median of 385 scans (interquartile range [IQR]: 314-429). The pooled area under the curve (AUC) as measure for the ability to differentiate BCC from non-BCC lesions was 0.852 (95% confidence interval [CI]: 0.833-0.870). Pooled specificity and sensitivity for a high-confidence diagnosis were 95.4% (95% CI: 93.2-96.9) and 31.1% (95% CI: 24.2-39.0), respectively.

CUSUM-integrated e-learning was successfully applied to train healthcare professionals to differentiate BCC from non-BCC on OCT. Trainees achieved a diagnostic accuracy suitable to start using OCT in clinical practice. This method can be applied to overcome the OCT assessor shortage and for training other medical skills.

Clinicaltrials.gov: NCT05634421.