COMPARATIVE ANALYSIS OF HYBRID MODELS FOR SKIN DISEASE DIAGNOSIS

ABSTRACT

The ever-growing prevalence of skin diseases, particularly skin cancer, necessitates the development of robust diagnostic tools. This project presents a Comparative Analysis of Hybrid Models for Skin Disease Diagnosis/Classification, focusing primarily on skin cancer. Employing an Adaptive Neuro-Fuzzy Inference System (ANFIS) as the core diagnostic framework, this study integrates two prominent machine learning optimization algorithms: Ant Colony Optimization (ACO) and Genetic Algorithm (GA). By amalgamating ANFIS with PSO and GA separately, we aim to enhance diagnostic accuracy and optimize model performance. To ascertain the efficacy of each hybrid model, a comprehensive comparison is conducted, evaluating their accuracy in diagnosing/classifying skin cancer. Three distinct feature selection techniques – Sequential Forward Selection (SFS), Correlationbased Feature Selection (CFS), and Recursive Feature Elimination (RFE) – are employed to enhance model robustness and discriminative power. Through rigorous experimentation and evaluation, this research endeavors to identify the most effective hybrid model for skin cancer diagnosis, thereby contributing to advancements in medical diagnostics and improving patient outcomes.