A Fiber-Optic Sensor-Embedded and Machine Learning Assisted Smart Helmet for Multi-Variable Blunt Force Impact...

Area of Science:

• Biomedical Engineering
• Materials Science
• Data Science

Background:

• Early diagnosis of mild traumatic brain injury (mTBI) is critical for effective clinical management.
• Current diagnostic methods lack detailed physical impact information.
• Smart sensors are needed for real-time monitoring of head impacts.

Purpose of the Study:

• To develop a smart helmet system for early on-site diagnosis of mTBI.
• To monitor complex blunt force impact events in real-time.
• To accurately recognize impact characteristics like magnitude, direction, and latitude.

Main Methods:

• Development of a smart helmet with an embedded fiber Bragg grating (FBG) sensor.
• Real-time monitoring of impact events in wired and wireless modes.
• Application of machine learning algorithms (e.g., S-SVM+, S-IBK+) to analyze FBG sensor data.
• Optimization of training datasets for machine learning models.

Main Results:

• The FBG sensor captures transient oscillatory signal 'fingerprints' reflecting helmet deformation.
• Machine learning models accurately identify impact magnitude, direction, and latitude.
• Boosted ML models demonstrated high accuracy with complex datasets.

Conclusions:

• The developed ML-FBG smart helmet system provides detailed physical information for mTBI diagnosis.
• This system enables quick and accurate recognition of transient impact events.
• The smart helmet shows potential as a crucial intervention tool for traumatic brain injury events.