<p>The prognosis of various malignant tumors, including breast cancer, can be effectively assessed using peripheral hematologic parameters obtained from patients. However, their significance is not examined across diverse breast cancer molecular subtypes. Detecting Triple-Negative Breast Cancer (TNBC) remains challenging because of its invasive nature, poor prognosis ability and high risk of distant metastasis. Immunohistochemistry (IHC) biomarker expression, clinical findings of the patients and morphological features derived from histopathology assessment are important for estimating the survival rate of breast cancer patients and are considered as significant components in diagnosis and prognosis. Recently, a two-step procedure has been suggested for diagnosing TNBC, including medical imaging and IHC, although this method can be time-consuming and operator-dependent. Therefore, the diagnostic efficiency of TNBC needs to be enhanced by adopting rapid and advanced technologies. In this research work, a new deep learning-based framework for TNBC detection and survival prediction is presented. At first, the required images to perform TNBC detection are collected from available databases. These images are used for the TNBC prediction process through Attention-embedded Adaptive Efficient Net, where several parameters are tuned using the Modified Mountaineering Team-Based Optimization to improve the TNBC detection and prediction effectiveness. Then, the risk and survival rate prediction is carried out by the Cascaded Artificial Neural Network, and the predicted survival rate is obtained. The proposed system has the potential to significantly improve the prediction of TNBC disease-affected outcomes. Finally, different experiments are carried out in developed TNBC detection and prediction models and also the effectiveness of the proposed scheme is contrasted with several baseline models to prove its enhanced performance.</p>