Android-Based Intelligent System for Early Lung Cancer Detection Using ML Techniques

Abstract

Lung cancer is one of the most prevailing reasons of cancer-related deaths in the globe and this is mainly because of the late diagnosis of the disease since at an early stage; the cancer may have no or limited symptoms. The early diagnosis is essential to enhance survival and treatment rates of patients, but the conventional diagnostic tools, including CT scans, X-rays, and biopsy procedures, are resource-intensive, expensive, and need special personnel, making their availability unavailable especially in remote or underserved regions. The proposed study build an Android-based smart application on early lung cancer detection with the help of machine learning procedures, integrating portability, efficiency, and diagnostic accuracy. The available data is used in the proposed system and includes publicly available data, such as lung CT scans, chest X-ray images, and related clinical data. Image normalization, noise reduction, and feature extraction also constituted preprocessing steps to improve the quality of the data and the model performance. The automated processing of medical images was done with Convolutional Neural Networks (CNN) and other structured clinical data were all processed using Support Vectors Machines (SVM) and Random Forest algorithms to enhance the classification accuracy. The trained models have been incorporated in an Android application in the form of TensorFlow Lite, and they are able to execute real-time inferences on mobile devices with a minimal computation burden. The findings proposed that the system has high levels of diagnostic performance, with an accuracy of between 90% and 96%, sensitivity of between 88% and 95%, specificity of between 89% and 95% and a balance of precision and recall. The proposed system will offer quick, dependable, and convenient early screening of lung cancer as opposed to currently used mobile health applications, as well as the conventional methods. These results show that machine learning used with mobile platforms can provide a scalable and feasible way to enhance timely diagnosis, assist healthcare professionals, and improve patient outcomes, especially in less-resourced and remote environments.