Early Lung Cancer Diagnosis Using a Hybrid CNN–BiLSTM Deep Learning Model: A Step Toward Precision Medicine

Lung cancer is regarded as one of the most lethal malignancies, with a 5-year survival rate that positions it among the three most fatal cancers worldwide. Successfully addressing lung cancer necessitates early identification for prompt tailored therapies. Nonetheless, guaranteeing early detection presents a significant difficulty, leading to the development of novel strategies. The advent of artificial intelligence(AI) provides groundbreaking options for lung cancer prediction. The necessity to improve artificial intelligence models remains a priority, especially in precision medicine, signifying a fundamental shift in healthcare. This article presents a hybrid deep learning methodology for lung cancer diagnosis utilizing patient medical records, combining Convolutional Neural Networks (CNN) with Bidirectional Long Short-Term Memory (BiLSTM) and Attention Networks. A comparative analysis utilizing the MIMIC IV dataset demonstrates the model's superiority, attaining a Matthews correlation coefficient (MCC) of 96.2% and an accuracy of 98.1%, surpassing LSTM and BioBERT, which achieved MCCs of 93.5% and 95.5% with corresponding accuracies of 97.0% and 98.0%, respectively. Our model greatly surpasses the compared models, demonstrating its exceptional performance and potential influence in the sector. This discovery represents a substantial advancement in the accurate and early identification of lung cancer, highlighting the continued need for the refining of AI models in precision medicine.