A Patient Response Prediction System using Multi-Scale Feature Fusion Semantic Segmentation Network and Med-YOLO

Nowadays, most of the people are getting affected with various health issues such as typhoid, diabetes and cancer. A human immune system plays a crucial role in maintaining and balancing human health and reduces the risk of getting ill. Immunotherapy is a treatment which utilizes human body immunity power to fight cancer. There are many challenges in the existing models such as imbalance data, heterogenic data, uncertainties, and lack of temporal data. These challenges are resolved by imposing deep learning techniques and generating a valid information with the clinical details of patient health conditions. Therefore, a hybrid model name called Multi-Scale Feature Fusion Semantic Segmentation Network (MSSNet) and You Look Only Once (MedYOLO) for feature extraction and detection. Initially, Synthetic Minority Over-sampling Technique (SMOTE) is employed to overcome the issue of imbalanced data; further MSSNet is imposed for extracting the optimal features and detected using MedYOLO. Also, to get detailed information and generate cohort, the data collected from GSC hospital with overall details that are essential to predict the patient response to immunotherapy in cancer treatment. In the end, MSSNet-MedYOLO obtained higher accuracy of 98.16% for an effective prediction of patient response which is better than existing Convolutional Neural Network (CNN).