EDGE ARTIFICIAL INTELLIGENCE FRAMEWORK FOR REAL-TIME BREAST CANCER DIAGNOSIS IN SMART HEALTHCARE

Abstract

One of the major factors responsible for the death due to cancers in women around the world is breast cancer, which calls for the importance of having intelligent diagnostic systems that could support decision making in healthcare applications. This research proposes Edge AI Framework for Real Time Detection of Breast Cancer in Smart Healthcare that uses machine learning algorithms and edge computing to provide real-time detection of cancer without any lag while preserving privacy issues. This framework exploits the Enhanced Breast Cancer Diagnostic dataset, which includes all the details about the morphological and clinical characteristics of the tumors to detect whether the tumor is benign or malignant. Data preprocessing methods like handling missing values, normalizing, and feature engineering are done initially. Thereafter, various machine learning models such as, but not limited to Random Forest, XGBoost, LightGBM, CatBoost, SVM, and Gradient Boosting models are built and evaluated to select the optimal prediction model. Selected model is optimized and deployed in the edge layer to ensure local prediction from the healthcare devices without the involvement of the cloud connection. This approach substantially increases the speed of the process, eliminates dependence on the network and its costs, and guarantees the security of patient data. To ensure transparency and confidence in the clinical decision-making process, Explainable AI (XAI) algorithms are applied for explaining predictions and selecting the most informative diagnostic features. Experimental results prove that edge-based AI model demonstrates outstanding diagnostic performance in terms of accuracy, precision, recall, F1-score, and AUC. The proposed system is more efficient compared to the cloud-based system since it performs better during operations. In addition, the proposed system is scalable with IoMT devices and smart healthcare systems, thereby enabling constant monitoring of patients and intelligent diagnosis. The proposed system can be used in order to detect breast cancer early through the provision of secure diagnostic services.