Early and proper diagnosis of cancer is one of the urgent problems of modern healthcare. This study suggests that an AI-based diagnostic system combines medical imaging with biomarker analysis using a pure experimental design technique. The imaging information was retrieved using Cancer Imaging Archive (TCIA), whereas the biomarker data was retrieved using the METABRIC database. Image resizing, normalisation, and augmentation of medical scans were performed in preprocessing, and, as the features of biomarkers, encoding was performed using Min-Max normalisation. This system uses a deep convolutional neural network, ResNet-50, to classify images and a Random Forest algorithm to classify tabular data of biomarkers. These two models were trained and tested separately with the help of Python-based frameworks such as TensorFlow, Keras, Scikit-learn, and the results were integrated through the soft voting ensemble. The models had a diagnostic reliability with an AUC of 1.00 and 90% accuracy, respectively, which means that the models are effective in their ability to provide diagnostic reliability. The above findings confirm the utility of incorporating deep and ensemble learning in multimodal classification of the cancer diagnosis, which is a potential clinical decision receiver and an early diagnosis tool.
Chizoba et al. (2026) studied this question.