PSYCH 403A1 — Assignment 6, Option C: Engineering a Consumer Wearable Brain Monitor

The wearable brain monitor uses a hybrid EEG-fNIRS system integrated into a lightweight headband. This design combines dry EEG electrodes to capture electrical neural activity with functional near-infrared spectroscopy for measuring hemodynamic responses. The hybrid approach overcomes the limitation that no single modality can represent the whole picture about brain function since different measurement techniques provide complementary information about neural processes. (Hong & Khan, 2017).

The system configuration includes 8 dry EEG electrodes placed at frontal (Fp1, Fp2, F3, F4) and parietal (P3, P4, Pz, Oz) locations, along with 4 fNIRS channels consisting of 2 source-detector pairs positioned over the prefrontal cortex bilaterally. This setup is focused on the prefrontal and frontal regions as they are the most important for executive function, attention, and cognitive control. The most relevant areas of consumer applications are in focus, stress, and cognitive performance markets.

It extracts the raw signals, four primary metrics, each based on established research. The Focus/Attention Score (0-100) is calculated based on frontal theta power, proxying cognitive control and sustained attention because frontal theta serves as a mechanism for cognitive control processes (Cavanagh & Frank, 2014). In combination with increases in oxygenated hemoglobin in the prefrontal cortex measured using fNIRS, it has been established that increased oxygenation in the prefrontal cortex is related with attentional demands and working memory load.

The Stress/Relaxation Index (0-100) is based on parietal alpha power, since it increases during relaxed wakefulness states. Alpha and theta oscillations have been shown to reflect cognitive and memory performance (Klimesch, 1999). These are combined with prefrontal hemodynamic stability metrics because stress responses alter cerebral blood flow patterns. Both electrical and hemodynamic measures reliably detect the sensitivity of cognitive task demands. fNIRS has demonstrated applications across a wide range of cognitive assessments (Ferrari & Quaresima, 2012).

To assess the quality of sleep in overnight mode, the device automatically detects sleep spindles, which are 12-15 Hz transient oscillations in parietal channels. Sleep spindles are really important and relate to memory consolidation, and decreased activity of sleep spindles has been shown in schizophrenia patients, reflecting the clinical importance of detection (Ferrarelli et al., 2007). Using automated classification based on the analysis of frequency bands and other features, the system classifies the various stages of sleep into Wake, Light Sleep, Deep Sleep, and REM. This hybrid approach offers validation when both the electrical and hemodynamic signals indicate the same cognitive state. The confidence in the metric goes up because hybrid brain computer interface techniques have shown different positive developments regarding the integration of different modalities that were mentioned as providing better classification results (Hong & Khan, 2017).

The device is designed in a headband style because it ensures stable sensor contact in the frontal and parietal regions, yet is lightweight and comfortable. Weighing in under 100 g, the device is made using medical-grade silicone, breathable mesh, and an adjustable strap, so it fits various head sizes and avoids creating pressure points for all-day comfort.

The battery provides about 12 h of usage and is recharged via USB-C. It has Bluetooth for wireless data transfer. Most signal processing occurs on board, while still allowing for instant streaming, and is designed to minimize reliance on a smartphone. The housing is also reasonably sweat resistant and durable for daily and light physical use.

The system is designed to be sweat resistant and robust, and all electronics feature a protective covering. The system can be worn over or under hair with little signal loss. The system has Bluetooth 5.2 Low Energy for wireless communication, and can be charged via USB-C. It is powered by a 250 mAh lithium-polymer battery that lasts up to 12 hours, fully charging in about 90 minutes, and stores onboard 16 GB memory which equates to approximately 7 days backup data. Most of the signal processing is done onboard, although data streams can go to a smartphone for visualization.

The EEG subsystem has 8 dry active electrodes with integrated pre-amplifiers (500 Hz per channel; 24-bit; and a bandwidth of 0.5-100 Hz). One of the key advantages of dry electrodes is that they do not require messy gel setups which enhances the proposed ease of use for a consumer system (Casson et al., 2010).

The fNIRS subsystem utilizes dual-wavelength LEDs (760 nm and 850 nm) paired with silicon photodiode detectors, sampling at 10 Hz. The fNIRS applies the Modified Beer-Lambert Law for estimating hemodynamic changes, and has standard artifact correction (Huppert et al., 2009). The continuous wave mode of operation keeps costs and power requirements low while adding reasonable sensitivity (Pinti et al., 2018).

To enable signal quality with hair, sweat, and movement, the device features spring loaded dry electrodes, silicone contact tips, and motion correction using accelerometers. Optical system components with ambient light shielding and noise reduction filtering are included in the design. Tradeoffs include 8 channels instead of 32–256 channels and a 10–15 dB lower SNR for dry electrodes; however, strong motion filtering and artifact rejection are in place.

A round Focus Score meter on the real-time dashboard of the smartphone app shows scores ranging from 0 to 100 and updates every 10 seconds. The backgrounds are color-coded for quick and easy visual feedback regarding the states: relaxed is identified with blue, focused with green, and stressed is shown as amber. There are frequency visualizations giving a 4-band, reduced resolution view of relative alpha, beta, theta, and gamma power. A graph that shows hemodynamic prefrontal oxygenation over the last 5 minutes is displayed next to a session timer.

There are multiple timescales for feedback: the continuous mode provides numerical scores and updated graphs every 10 seconds in real time, while a summary feature compiles automatic hourly summaries of peak focus periods and instances of stress based on individual detection of patterns for later consideration. Daily reports include the overnight period with analysis of sleep architecture including total sleep time, sleep efficiency, and spindle density measurements, and a weekly report using machine learning to provide insights and identify dimensions of broader patterns that correlate specific activity with general periods of optimal cognitive states.

It has on-device and cloud-based machine learning, to offer actionable guidance and support. After seven days of baseline recordings, it specifies separate individual patterns of brain activity that also respect individual variances from baseline levels. Pattern recognition algorithms make correlations temporally attached to analyze changing patterns relating to brain activity patterns, variable patterns that co-appear from baseline sleep, attention, and stress metrics. It employs both on-device and cloud-based machine learning to provide actionable recommendations.

After 7 days of baseline recording, the program establishes individual specific norms for focus, stress, and sleep measures that take into account their individual baseline brain activity variability. Subsequently, through algorithm pattern recognition, it establishes correlations for time-varying cognitive performance states or predictable behavioral factors related to levels of stress. The recognition system will suggest activities aligned with the needs of the brain based on current brain states; for example, the system will recommend breathing practices when alpha activity is too low or suggest taking a break when the focus measures are declining.

Over time, individualized insights evolve as the system learns individual patterns and can recommend the optimal time to work and add activities that improve or impair cognitive performance. The primary use case is in real-time focus optimization for knowledge workers, students, and creative professionals. Users can use the continuous feedback to identify when their cognitive performance behavior is optimal for conducting deep thinking work, detect when their attention drifts before some kind of productivity collapse, and confirm whether focus techniques such as Pomodoro intervals or meditation practices actually are effective.

Secondary uses include meditation training using instant alpha feedback that can confirm practice is having an effect and provide objective feedback for measuring states of relaxation that go above and beyond what self-report states provide. Sleep optimization happens through spindle tracking that can help improve sleep hygiene and stress management benefits from early detection. Cognitive health monitoring through long-term tracking can detect subtle changes that may warrant professional evaluation. This proves valuable for people and professionals to wear and ensure they have accurate data to meet their needs continually.

In order to ensure the scientific credibility of the device and contend with the weaknesses of previous systems, we propose a three-phase validation approach.

Phase 1: Laboratory validation, using approximately six months (the timing will depend on participant engagement) with 50 participants who will wear the device (alongside concurrent collection of standard 128-channel EEG and research-grade fNIRS data). To explore the validity of the device, we will measure the correlation of device metrics with established cognitive task performance metrics.

Phase 2: Ecological validation will take a 12-month period with 500 end users of the device using the device in the field in various settings (home, office, school). We will use the validation process to measure the correlation of device metrics with objective behavioral performance metrics (typing speed, typing accuracy, timeliness for task completion), subjective rating measures, and standardized mood measures. Where applicable, we will also extend investigations to important behavioral markers such as how often participants pick up their phone, and how often work contexts are changed (i.e., switching between more than 2 applications). The results will be submitted to a peer-reviewed journal, and the data available for public use.

Phase 3: Clinical validation in partnership with sleep clinics over approximately 18 months to directly validate the sleep staging metrics to polysomnography gold standard measures and to compare focus metrics performance measures from ADHD populations to clinical assessments with the goal of demonstrating >80% agreement with gold standard clinical assessments.

In the interest of not making misleading claims, it is entirely transparent by openly sharing all validation data, showing real-time data quality scores in the app, and using explicit uncertainty language (i.e., "this signal has 99% confidence" and descriptive language for considerations such as hair interference). There are no claims being made for capabilities that have not been validated. The device is available for independent and external research verifications. The regulatory strategy will first position the device as a general wellness device (not making medical claims). An independent scientific advisory board of leading EEG analysis, fNIRS methodology, and sleep research scholars would convene to conduct reviews of the algorithm's validity, advance validation protocols, co-author peer-reviewed publications, and ultimately ensure no scientific overreach occurred and the validity assessments were independent.

The cost breakdown of the electronics is based on a projected cost of 10,000 units: EEG electronics are $28, which includes the 8-channel ADC and active electrodes; fNIRS elements are at $22, which covers all LEDs, photodetectors, and drivers; processing is $18, which covers the Bluetooth module; battery and charging circuitry is under $8; mechanical components and materials are at $12; and assembly labor is $7 for total manufacturing costs of $95. The $199 price point is pitched competitively with the Muse S ($250), with a price advantage over premium wearables in the market. This price point offers attractive profit margins and can provide funding for product research and development, marketing, and customer support. The price also gives an accessible opportunity for a subscription at $19.99/month for access to advanced AI summary analytics, but does not affect core functionality. We want to ease users into a subscription model that is financially sustainable.

The target audience is productivity for professionals, including software engineers, writers, as well as meditation practitioners who want objective feedback to practice meditation, and students who want to optimize their study sessions. Compared to current competition, Muse S is $250 and provides 4-channel EEG; Neurosity Crown is $799 for 8-channel EEG for developers; Kernel Flow is not available to consumers; while Fitbit and Oura only track sleep and provide no cognitive metrics. The proposed device will include differentiation in that it will be the only hybrid EEG-fNIRS to market under $500 and have the best 12-hour battery life to compete with productivity optimization.

Additionally, robust privacy and ethical safeguards are in place. This means entire raw EEG and fNIRS neurophysiological data is end-to-end encrypted; with on-device processing for real-time metrics, data will never leave the phone without explicit consent from the user. Optional cloud synchronization entails the use of anonymized data, and at each step, users retain full control over sharing and deleting data. Ethical safeguards encompass terms of service to avoid forced use by employers in workplace settings to deter coercion, data portability so users can export all raw data for personal use or independent analysis. Transparent limitation communication involves a highly visible "This is not a medical device" disclaimer in the application itself, while research participation is voluntary and opt-in only for users. Misuse prevention entails software detection that alerts the user to a workflow that is impossible to perform 24/7, educational few steps explaining healthy brain optimization versus obsessive tracking behaviors, and collaboration with mental health organizations to ensure there is accountable marketing.