Automated Detection of Lung Cancer Levels Based on Patient’s Lifestyle Information

Lung cancer begins in the lungs, where cells grow out of control to create tumors. Lung cancer is one of the major causes of cancer-related deaths globally. With less than 20% of cases surviving for five years, prompt and precise detection is essential. This study discusses the shortcomings of the existing medical screening protocols as well as the length of time needed to discover lung cancer stages. The project aims to create an automated system that uses lifestyle data from cancer patients to accurately identify the cancer level, reducing the possibility of human mistakes and speeding up the diagnostic procedure. Based on answers to lifestyle health questions like diet, obesity, smoking habits, passive smoking, chest pain, and bloody coughing, the algorithm predicts the level of lung cancer. To build the prediction system, two machine learning classifiers are used with a dataset containing medical records of 1,000 patients who have lung cancer. The approach involved loading data from an Excel file into MATLAB and using cross-validation for precise model creation. The process is followed by constructing a confusion matrix and then extracting performance metrics from it including accuracy, recall, and F1-score. The first model achieved a classification accuracy of 96% using the Support Vector Machine classifier, which was further improved to 100% using the Error-Correcting Output Codes (ECOC) model. Additionally, a Lung Cancer Level Prediction web application is developed using "Streamlit" to assist physicians in predicting lung cancer levels for any new patient faster and cheaper than the regular methods. Predictive analytics is practically used in healthcare by this application, which lets the user choose one of several machine learning models to deliver predictions engagingly and interactively. This web-based application improves usability and accessibility for both patients and healthcare providers.