Lung and liver cancers present difficulties for earlier diagnosis because traditional diagnostic imaging procedures, biopsies, and the steps to obtain them are invasive and costly. Breath-based diagnostics represents an approach that is non-invasive, by examining volatile organic compounds (VOCs) that indicate changes in normal metabolic processes associated with cancer. This chapter presents a hybrid electronic nose (e-nose) system that uses machine learning algorithms, Metal Oxide Semiconductor (MOS), and Quartz Crystal microbalance (QCM) sensors to detect lung and liver cancer. The Random Forest machine learning algorithm, Support Vector Machine (SVM), and Weighted Discriminative Extreme Learning Machine (WDELM) models are used in an e-nose to classify, reduce dimensionality, and condition the sensor signal for analysis and processing. The design of e-nose provides a scalable and portable method of accurately screening for cancer in real-time in clinical and community settings. The primary contribution of this work is the design of a hybrid E-nose system based on a combination of MOS and QCM sensors interfaced with machine learning algorithms (RF, SVM, and WDELM) that ensure a high level of accuracy, of a non-invasive approach to cancer detection.