Artificial Intelligence-Assisted Dermatologic Screening: Epidemiology and Clinical Features of Basal Cell Carcinoma,...

This literature review synthesizes contemporary evidence regarding the epidemiology, screening guidelines, clinical manifestations, and machine-learning solutions for four prevalent non-melanoma skin lesions: basal cell carcinoma (BCC), squamous cell carcinoma (SCC), seborrheic keratosis (SK), and actinic keratosis (AK). This study presents a summary of common indices and recent screening alternatives, accompanied by a critical assessment of contemporary advancements in artificial intelligence (AI) and machine learning (ML) for the identification and classification of images utilizing standardized benchmark databases. The literature search and selection focused on peer-reviewed studies published from 2018 to December 2024, emphasizing diagnostic performance, datasets, preprocessing methodologies, and assessment metrics. This work compares and contextualizes reported results, highlighting the challenges posed by different study designs and biases in datasets that hinder direct comparisons among studies. The consistency of deep learning classifiers in lesion detection, the significance of sensitivity-oriented thresholding for early detection applications, and challenges associated with class imbalance and the under-representation of darker skin tones in publicly accessible datasets are studied. With practical implications for clinical adoption, emphasizing targeted screening of at-risk populations, the supplementary benefits of dermoscopy and the imperative for multi-center, demographically diverse validation have been concluded. Additionally, future research on standardized reporting, external validation, and interpretable, workflow-compatible AI systems has been proposed.