Artificial Intelligence (AI) is transforming healthcare by improving diagnostic tools. One significant application is detecting skin cancer. In our recent study, we explored how boosting algorithms can enhance classification accuracy for identifying skin cancer types using the PAD UFES 20 dataset.

We started by building a Convolutional Neural Network (CNN) feature extractor to process the skin cancer images. The CNN helped identify important patterns in the images, which were then refined using data preprocessing techniques like Principal Component Analysis (PCA) to simplify the data without losing key information.

Next, we trained three powerful boosting algorithms—CatBoost, XGBoost, and LightGBM (LGBM)—to classify the extracted features. Among these, XGBoost delivered the highest accuracy, reaching 73.2%. This result highlights how advanced machine learning techniques can improve diagnostic precision.

Finally, we analyzed the features that contributed the most to the model’s predictions. Understanding these features provides valuable insights into how AI systems interpret medical data, helping researchers and healthcare professionals build more reliable tools.

Our work showcases how AI is shaping the future of healthcare, offering faster and more accurate diagnostic methods for skin cancer. With further research, these technologies can support early detection and better patient outcomes worldwide.

The research paper will be available soon on MDPI Engineering Proceedings.