Skip to main content
2024 BMES Annual Meeting Main banner
Icon Legend
This session is not in your schedule.
This session is in your schedule. Click again to remove it.

Device Technologies and Biomedical Robotics

Device Technologies and Biomedical Robotics - Poster Session C

Poster J1 - Niura: A Modeled System with In-Ear Silicone EEG Electrodes Embedded in Wireless Earbuds for Productivity Tracking through FFT and PCA Analysis

Friday, October 25, 2024
10:00 AM - 11:00 AM EST
Location: Exhibit Hall E, F & G

Introduction:  With over 15 million college students, the United States has seen a surge in those pursuing higher education in recent years. However, many of these students suffer from procrastination and poor studying habits, phenomena that significantly contribute to an aggregate reduction in productivity levels. Detecting these drops in productivity and offering innovative and effective solutions to them is critical in boosting overall productivity and creating healthier study habits. Despite this burgeoning issue, there has been no cost-effective and robust solution to these problems, especially one that factors in students’ mental health and wellbeing. Therefore, a low-cost, non-invasive device that captures electroencephalography and provides real-time suggestions based on the data is necessary. The Niura earbud, which integrates conductive silicone electrode technology with an earpiece, uses modified electrocardiogram analog front-end circuitry to detect electrical differences in neural signaling. By applying a Fast Fourier Transform, it can identify and display frequencies in real time on a digital EEG monitor. These frequencies are extracted from complex sine waves. Ultimately, with Niura, students will have a more comprehensive understanding of their brain health and have a much larger incentive to be productive, potentially leading to long-term changes in student and work culture in the United States and around the world.

Materials and

Methods:  The following devices were integrated into a verifiable prototype. An earbud device in which the main processor, nRF5340, operates in high voltage mode, powered by either the battery or 5V input power. A power section in which the power multiplexer, TPS2113ADRBR, switches between battery and charger power and avoids digital noise coupling into analog power. The primary device EEG device, ADS1291IPBS, is a chip powered by 2 LDO outputs that communicate through SPI with the nRF5340. An audio amplifier, TLV320DAC3100, from Texas Instruments is powered by the digital power input, while analog power is generated internally by an integrated LDO. The EEG electrodes set in the earbud tips are positioned in the ear canal to make contact with the skin covering the skull's temporal bone. A low or high-power, low-noise printed circuit board (PCB) with an analog front-end (AFE) measures brain electrical activity. Artificial intelligence interprets collected data and can transmit it to other wired or wirelessly connected devices.

For computer-aided design (CAD) modeling, a combination of Blender and Solidworks was utilized for appropriate models. Polylactic acid (PLA) filaments were used for the majority of prototype printing. A combination of CURA and AutoMaker Bot machinery was used to print prototypes at a designated infill density of 20% for optimal wear. 30 shore liquid silicone electrodes are used for comfort and compatibility with carbon fiber and graphite elements, with integrated 281 gr/m2 graphite & C80/44 Conductive Graphite for EEG signal detection.

Results, Conclusions, and Discussions: The proposed ML algorithm integrates pre-existing 2-channel EEG database with an in-house CSV database from Muse headsets to form the baseline dataset. Employing PCA facilitates the extraction of key EEG information while maintaining its integrity and reducing processing complexity. K-fold cross-validation ensures the models' generalizability and guards against overfitting.
The brain's electrical activity in EEGs are divided into clinically-established frequency bands: alpha, beta, theta, delta, and gamma. Increased theta activity in the temporal region suggest a decline in attention and alertness. This pattern is associated with states of relaxation or drowsiness, while gamma activity is associated with memory formation and problem-solving2. Prototype testing of the conductive silicone electrodes and past iterations of the circuit led to readings of the topographical mid-frontal and temporal lobes with detection of Beta, Theta, Gamma, and Sub-Alpha waves at approximately 4400 signals per minute. The frequencies measured vary from 3.9 Hz to 50.78 Hz, showcasing that the Niura earbuds may differentiate between various EEG signals. From studying these signals and implementing machine-learning software such as PCA, insight is gained into active cognitive processing, variations in attention, and greater focus. Delta waves were not recorded during tests since delta waves can only be detected during deep-sleep sessions. Proposed electrodes show industry standard levels of electrical resistance, roughly 6-8 ohms, and the earbud prototype detects the brain’s major frequencies. Past published studies have shown approximately 82% efficacy in continuous detection of attentiveness and rest states through EEG signals omitted from the frontal and temporal regions of the brain.

Acknowledgements (Optional):  We would like to thank the New York Institute of Technology with providing us the relevant lab space to undergo this project. Thank you to NSF I-Corps for fueling the CDI process of this project. Thank you to Michigan ZLI and NYU Stern for unwavering support regarding the entrepreneurial components of this project.