Chatbot Enable Brain Cancer Prediction Using Convolutional Neural Network for Smart Healthcare

Abstract

Recent advancements in deep learning have shown impressive accuracy in detecting cancer from images and human radiologists and gaining suitability for clinical use. However, a key challenge arises from the inherently unexplainable nature of these unsupervised models, hindering their clinical adoption due to the lack of transparency. Regulatory concerns also discourage the application of unexplainable models in clinical settings, emphasizing the need for interpretability. To tackle these challenges, recent research focuses on adapting deep learning architectures to enhance explainability in cancer detection from magnetic resonance imaging (MRI) snapshots. Natural language processing (NLP) techniques are harnessed to create a smartphone chatbot using convolutional neural networks (CNNs). This system employs Bot application programming interfaces (APIs) for machine integration and SMS subscriber connectivity for smart health care in smart cities. Utilizing CNNs, the chatbot predicts diseases based on user-input symptoms identified through NLP, incorporating medical ontologies to enrich its knowledge base. The CNN achieves an impressive testing accuracy of 98.23%, surpassing the artificial neural network (ANN) at 82.63%, with a precision of 95% compared to the ANN at 80%. This underscores the CNN’s superiority in accurately predicting brain cancer, highlighting its potential in clinical applications.