Lung Cancer Relapse Prediction Using Self-Organizing Maps: A Clustering-Based Approach for Early Detection

Lung cancer relapse is very often found to affect patient prognosis and treatment outcomes potentially and may lead to more challenging management and poorer survival rates. Early prediction of relapse is necessary for timely intervention and personalized treatment plans and to handle the key effects of relapse on prognosis and treatment, such as prognosis and survival, treatment outcomes, impact on treatment strategies, psychosocial effects, and potential for clinical trials. With this motivation, this paper proposed Self-Organizing Maps (SOMs) to predict lung cancer relapse. The SOM technique is considered for predicting lung cancer relapse due to its dimensionality reduction, ability to recognize and cluster patterns, visual way to interpret complex medical data operation without the need for labelled training data, capability to handle non-linear relationships, adaptability to evolving data, and noise tolerance. It is investigated that this unsupervised learning technique clusters patients based on their clinical and pathological features, providing a valuable tool for identifying relapse-prone patients. The results reflect that the proposed approach attains high prediction accuracy by leveraging the clustering capabilities of SOMs and effectively classifying patients into relapse and non-relapse categories. It can be concluded that the SOMs have the potential to enhance early detection and personalized treatment of lung cancer relapse.