Efficient Edge AI for Neonatal Care: Implementing HIE Grading on Snapdragon Edge Device.

Area of Science:

• Neuroscience
• Artificial Intelligence
• Medical Technology

Background:

• Hypoxic-Ischemic Encephalopathy (HIE) requires timely diagnosis within six hours for effective treatment.
• Current HIE diagnostic methods are complex, time-consuming, and require specialized expertise.
• Accurate HIE severity grading is crucial for optimal clinical management.

Purpose of the Study:

• To develop and implement an automated system for grading Hypoxic-Ischemic Encephalopathy (HIE) severity.
• To provide clinicians with an accessible and secure tool for real-time HIE diagnosis and decision support.
• To enhance the efficiency and accuracy of HIE assessment in clinical settings.

Main Methods:

• EEG signals were converted to the audio domain.
• Spectrogram representations were classified using a 2D Convolutional Neural Network (CNN) with 8-bit integer quantization.
• The system was implemented on a Qualcomm RB3 Gen 2 edge device for real-time analysis.

Main Results:

• The automated system achieved high accuracy in differentiating mild and moderate EEG abnormalities.
• The system demonstrated robust performance suitable for neonatal intensive care settings.
• Achieved a fast inference time of 62 ms for practical clinical application.

Conclusions:

• The developed automated HIE grading system offers precise and timely decision support for clinicians.
• Implementation on an edge device enables real-time analysis and mobile accessibility.
• This AI-driven approach advances HIE diagnosis, improving patient care in neonatal intensive care units.