In recent instances, the exponential expansion of big data upswing (DU) has proven to be a powerful method of protecting against cyber security risks. Traditional data security techniques frequently fail owing to the complexity and volume of medical data, revealing possible vulnerabilities. This research provides a new technique for improving hospital data security utilizing unpolarized region-based Convolutional Neural Networks (R-CNN). The suggested Unpolarized R-CNN framework is capable of processing high-dimensional medical data, including Electronic Health Records (EHRs), medical imaging, genetic data, and sensor data, thanks to enhanced feature extraction and anomaly detection approaches. The integration of Region Proposal Networks (RPN) with unpolarized neural network layers is critical to this technique, as it allows for the impartial detection and classification of data breaches and harmful activity. The unpolarized design ensures that all data points are analyzed equally, which improves the identification of tiny irregularities that indicate cyber security problems. The framework also incorporates a real-Time intrusion detection system that monitors data flow, detects unwanted access, and ensures data integrity. The Unpolarized R-CNN performs better in detecting complex cyber threats and reducing false positives, which is critical for preserving trust in digital health systems. This technique also allows for secure data processing and storage, by health data privacy standards such as HIPAA. Experimental results demonstrate the framework's usefulness in a hospital setting, highlighting its potential to improve data security in healthcare with an accuracy of less error of 3% and an F1 Score of 92 FPR with 93% in unpolarized method. © 2025 Elsevier B.V., All rights reserved.