AI-Augmented Fetal Health Monitoring: A Cloud-based Secure Deep Learning Approach

Fetal health monitoring is a critical aspect of prenatal care, requiring accurate and timely diagnosis to ensure optimal outcomes. This study introduces SecureCloud-FHMNet, an AI-augmented, cloud-integrated deep learning framework designed to enhance the diagnosis of fetal conditions using cardiotocography (CTG) signals. The proposed model uniquely combines CNN for spatial feature extraction and attention-based BiLSTM for capturing temporal dependencies. Extensive experiments were conducted on a benchmark CTG dataset comprising 2126 records, with classification performed on both 3-class (Normal, Suspect, Pathologic) and 10-class morphological labels. The proposed model achieved an accuracy of 96.2% for 3-class classification and 93.5% for 10-class classification, outperforming traditional machine learning models like SVM and MLP. Hyperparameter tuning and ablation studies confirm the significant impact of attention and fusion layers on performance. Additionally, the cloud-based implementation ensures secure, real-time access and processing of patient data, enhancing usability in clinical settings. Visual analysis using confusion matrices and ROC curves further validated the model's predictive efficiency. This framework not only addresses diagnostic accuracy but also ensures data privacy and system scalability, paving the way for next-generation prenatal monitoring systems.