A Generative Adversarial Network (GAN) Technique for Internet of Medical Things Data

Area of Science:

• Medical Informatics
• Artificial Intelligence in Healthcare
• Biomedical Data Science

Background:

• Machine learning (ML) and artificial intelligence (AI) are increasingly used in healthcare for disease identification, patient monitoring, and clinical decision support.
• The Internet of Medical Things (IoMT) enables continuous remote patient monitoring, but data can be insufficient due to real-world issues like connectivity loss or poor adherence.
• Data augmentation techniques are crucial for generating sufficient synthetic datasets to train robust ML models.

Purpose of the Study:

• To apply Generative Adversarial Networks (GANs) for data augmentation in Chronic Obstructive Pulmonary Disease (COPD) monitoring using IoMT sensor data.
• To validate the accuracy and comparability of GAN-generated synthetic data against real-world patient data.
• To utilize explainable AI (XAI) for assessing the quality of synthetic data in the context of COPD monitoring.

Main Methods:

• Implementation of Generative Adversarial Networks (GANs) for synthetic data generation from IoMT sensor data.
• Collection and utilization of real-world patient data for Chronic Obstructive Pulmonary Disease (COPD) monitoring.
• Application of explainable AI (XAI) algorithms to compare synthetic and real datasets.

Main Results:

• Synthetic datasets generated by a well-structured GAN were found to be comparable to real patient datasets.
• The accuracy of the synthetic data was validated against real data using machine learning-based approaches.
• Explainable AI confirmed the reliability of the augmented data for ML model training.

Conclusions:

• GAN-based data augmentation is a viable method for creating high-quality synthetic datasets for IoMT-based COPD monitoring.
• The generated synthetic data can effectively supplement real-world data, improving the robustness of ML models.
• This approach offers a promising solution for overcoming data scarcity challenges in remote patient monitoring applications.