Pancreatic Cancer Detection Using Machine Learning
Miss 1 Jyothika K R, 2 Vijay Kumar N
2Student,4th Semester MCA, Department of MCA, BIET, Davanagere
1 Assistant Professor, Department of MCA, BIET, Davanagere
ABSTRACT
Pancreatic cancer is regarded as one of the most deadly types of cancer, primarily due to its often late diagnosis, swift progression, and absence of early symptoms. Timely and precise detection can significantly enhance patient survival rates. This initiative introduces a machine learning-based detection model intended to aid medical professionals in recognizing pancreatic cancer at its initial stages by utilizing clinical, genetic, and imaging data. The system employs robust machine learning algorithms, including Support Vector Machines (SVM), Random Forest, Logistic Regression, and XGBoost, which are trained on datasets featuring attributes such as age, gender, family history, genetic mutations, biochemical markers, and medical imaging indicators. The model pipeline encompasses data cleaning, feature selection, normalization, model training, evaluation, and deployment. To enhance the reliability of predictions, ensemble models and cross-validation techniques are utilized. The system's performance is assessed using standard medical diagnostic metrics, including accuracy, sensitivity (recall), specificity, precision, and ROC-AUC score. Sophisticated feature engineering is applied to
identify biomarkers that exhibit a strong correlation with the presence of pancreatic cancer. A user-friendly interface is created to enable healthcare practitioners to enter patient data and receive real-time cancer risk predictions. The predictions are illustrated with clear graphs and diagnostic indicators to facilitate clinical decision-making. This system is not intended to replace physicians but rather to support them in early screening, minimizing false negatives, and prioritizing high-risk patients for additional testing. By merging artificial intelligence with healthcare, this project advances intelligent diagnostic tools, precision medicine, and enhanced cancer care outcomes.
Keywords: Pancreatic cancer, early detection, machine learning, Support Vector Machine (SVM), Random Forest, Logistic Regression, XGBoost, clinical data analysis, genetic biomarkers, medical imaging, feature selection, data preprocessing, SHAP, LIME, diagnostic system.
