Machine learning based colorectal cancer risk assessment : A comprehensive framework for heterogeneous data integration

AI and machine learning branch have revolutionized the medical sector, demonstrating significant success in the diagnosis and treatment of various cancers, including colorectal cancer (CRC). The potential of machine learning models extends to personalized cancer care, poised to redefine the landscape of future medical practices. Early detection of colon cancer and polyps play a pivotal role in reducing the mortality and morbidity rates associated with colorectal cancer. This prospective study endeavors to establish a straightforward, valuable, efficient, and dependable scoring system for assessing the risk levels (low, medium, high) of colorectal cancer. This study uses heterogeneous parameters like age, gender, tumor stage, tumor grade and CEA. Pertinent score is assigned in the process of integration of these parameters to assess the risk level. In the first phase of the study the CT scan images were used to identify the tumor stage using neural network model and create a comprehensive dataset integrated with other parameters in the next phase. Utilizing this dataset, we implemented random forest model to demonstrate how these factors collectively participate in the risk assessment of colorectal cancer which further assists clinicians in diagnosis, treatment, and monitoring.