Sign in Subscribe
Machine learning

Breakthrough in Big Data: Researchers from Zagazig University Explore Differential Privacy in Machine Learning for Heartbeat Detection

Strategic Insights

— 1 min read

Researchers from Zagazig University have made significant strides in the field of big data by exploring the application of differential privacy in machine learning techniques for heartbeat detection. According to a newly published report, the team used deep learning models, including CNN, LSTM, GRU, and RNN, to classify heartbeat abnormalities while preserving patient data privacy. The study demonstrated that the GRU model achieved the highest accuracy of 99.5% in heartbeat abnormality detection, while also ensuring that individual patient data remained anonymous. This breakthrough has practical implications for real-world healthcare scenarios, where sensitive health data must be protected.

Key Takeaways:

  • The study utilized machine learning techniques in healthcare applications to achieve rapid growth and improved accuracy in heartbeat abnormality detection.
  • Researchers from Zagazig University developed a differential privacy approach to protect patient data while maintaining high accuracy in heartbeat abnormality detection.
  • The GRU model demonstrated the highest accuracy of 99.5% in heartbeat abnormality detection, followed by CNN (99.12%), LSTM (98.89%), and RNN (79.60%).
  • The study showcased the application of various deep learning models to classify heartbeat abnormalities through the incorporation of noise reduction filters, heartbeat segmentation, and resampling.
  • The findings of this study provide practical guidance for selecting effective and privacy-preserving deep learning models for heartbeat abnormality detection in real-world healthcare scenarios.
  • The research was conducted by Osama M. ElKomy, Ehab Rushdy, Sohaila Nasser, and Marwa M. Khashaba from the Department of Information Technology at Zagazig University.

Statistics:

  • The accuracy of the GRU model in heartbeat abnormality detection was 99.5%.
  • The accuracy of the CNN model in heartbeat abnormality detection was 99.12%.
  • The accuracy of the LSTM model in heartbeat abnormality detection was 98.89%.
  • The accuracy of the RNN model in heartbeat abnormality detection was 79.60%.
  • The study utilized the multimodal MIT-BIH polysomnographic dataset.
  • The researchers trained and evaluated the deep learning models on a total of 1,000 heartbeat samples.
  • The study achieved a high accuracy in heartbeat abnormality detection while preserving patient data privacy through the incorporation of differential privacy.

Sources:

  • Exploring differential privacy in CNNs, LSTMs, GRUs, and RNNs for heartbeat detection from multimodal data. Journal of Big Data, 2025, 12(1):1-17. (Journal of Big Data - https://journalofbigdata.springeropen.com)
  • NewsRx. Researchers from Zagazig University Describe Findings in Big Data (Exploring differential privacy in CNNs, LSTMs, GRUs, and RNNs for heartbeat detection from multimodal data). Information Technology Newsweekly. October 14, 2025; p 1039.

Read more