A Systematic Comparative Study on the use of Machine Learning Techniques to Predict Lung Cancer and its Metastasis to the Liver: LCLM-Predictor Model

Abstract

Listen

Translate article

Lung cancer is one of the major causes of cancer deaths with thousands of affected patients who have developed liver metastasis, complicating the treatment and further prognosis. Early predictions of lung cancer and metastasis may greatly improve patient outcomes since clinical interventions will be instituted in time. This paper compares the performance of different machine learning models including Decision Tree Classifiers, Logistic Regression, Naïve Bayes, K-Nearest Neighbors, Support Vector Machines and Gaussian Mixture Models toward the best set of techniques for prediction. The applied dataset includes various clinical features, such as respiratory symptoms and biochemical markers, for the development of stronger predictive performance. The models were cross-validated using testing and validation techniques aimed at generalizing the whole model with reliability in generating both train and test data. The results of the generated models are gauged using metrics of accuracy, precision, recall, F1-score, and area under ROC curve. Results obtained have revealed that the Decision Tree and KNN models also showed stronger predictive accuracy and strong classification performance, especially in early-stage lung cancer and liver metastasis. The present study is a comparison of the Decision Tree and KNN models, which hence denotes the potential of these models in clinical decision-making and suggests application to the development of diagnostic tools for the early detection of cancer. This provides a very useful guide that is applicable in the use of machine learning in oncology and helps pave the way to future research which would be focused on model optimization and integration into healthcare systems that would produce better management of patients and better survival rates.