Artificial Intelligence-Enhanced Optimization of Wireless Breath Sensor Arrays for Detection of Lung Cancer Using Fuzzy Logic-Guided Genetic Algorithm and Multimodal Machine Learning

Early detection of lung cancer remains critical for improving patient survival, yet current imaging-based screening methods are costly, invasive, and limited in accessibility. Here, we present a fully integrated wireless breath sensing platform that combines nanostructured chemiresistive (NC) sensor arrays with an AI-driven Fuzzy logic-guided Genetic Algorithm (Fuzzy-GA) for optimized volatile organic compound (VOC) detection. The sensor array features nanoparticle structured interfaces, enabling selective VOC adsorption to generate unique breath patterns. Data are captured via a portable low-current multichannel electronics module with real-time wireless transmission. Fuzzy-GA optimization identifies the most informative sensors, reducing array size while maintaining high diagnostic performance. Breath samples from lung cancer patients (n = 35) and non-cancer participants (n = 47) were analyzed using multiple supervised machine learning models (KNN, SVM, Random Forest, XGBoost, and CNN). This represents the first application of Fuzzy-GA to optimize breath sensor arrays. The optimized system, validated using breath samples from lung cancer patients and non-lung cancer controls, achieved high classification accuracy (up to 96%) with reduced system complexity, lower cost, and improved scalability for real-world deployment. The platform offers a clinically viable, non-invasive diagnostic tool with potential for at-home monitoring and broader disease detection.