i-Cognia: Brain-Controlled System Empowering Independent Living for People with Motor Impairments

The Challenge

Over 2.5 million people live with severe motor impairments such as ALS and quadriplegia.

For many, an eye blink is the only reliable voluntary movement.

Around 1 in 400 individuals become fully dependent on caregivers despite full cognitive awareness.

The problem is not cognition, but the loss of a practical output channel for communication and control.

Impact

This loss of motor function creates major human and economic challenges:

Calling for help, sending messages, or operating household devices becomes extremely difficult without assistance.

Constant dependence on caregivers reduces independence, privacy, and quality of life.

Limited communication can increase frustration, isolation, and safety risks in everyday situations.

Lifetime care expenses can reach several million dollars per person.

Limits of Current Systems

Existing assistive systems are often:

Expensive, making them inaccessible for many families.

Invasive, requiring surgery or complex medical intervention.

Unreliable for everyday use outside controlled settings.

Most systems need specialized clinical setups, perfect conditions, and movement that some users no longer have. As a result, these systems fail to provide practical, consistent support in real environments.

Most systems don’t rely solely on eye-blinks, often only remaining ability.

What Is Needed

An effective assistive solution must be:

Low-cost and accessible.

Non-invasive.

Reliable in real-world settings.

Easy to use at home.

Suitable for people with extremely limited mobility.

Able to restore communication and environmental control through a simple intentional action, such as an eye blink.

Such a solution is essential to preserve independence, dignity, safety, and quality of life.

Background & Concept Development:

Considering the requirement of using blinks for control, I began by developing a deep understanding of how eye blinks work and how to capture the distinct blink state as an independent communication channel.

CRP (Corneo-Retinal Potential): The eye acts as an electrical dipole; blinking perturbs the surrounding electric field, producing a high-amplitude, low-frequency transient at frontal EEG sites (Fp1/Fp2).

Hypothesis: Blink-induced signals can be used for intentional control.

Challenges:

Initial experiments revealed several challenges in reliable blink detection:

Separating voluntary vs. involuntary blinks, especially during natural behaviors like speaking, facial movement, or fatigue

Variability in blink amplitude and baseline noise across users, electrode placement, and session conditions

Presence of motion artifacts and environmental noise, which can mimic or distort blink signals

Maintaining real-time responsiveness while minimizing false positives and missed detections

Eye-Blink Algorithm Overview:

i-Cognia’s blink detection algorithm uses a three-phase calibration + real-time inference pipeline to ensure accuracy and robustness. It achieves high performance through:

Adaptive baseline (sliding-window normalization): Detects relative deviations instead of fixed amplitudes

Robust estimation (Median + MAD): Reduces sensitivity to noise and outliers

Personalized thresholds: Ensures consistency across users and sessions

Eye-Blink Algorithm Steps:

Adaptive Calibration:

Because blink amplitude and baseline noise vary, voluntary blinks must be distinguished from non-intentional activity. I designed an inference pipeline based on adaptive signal statistics rather than fixed thresholds:

Phase 1: Estimate baseline noise using median and MAD

Phase 2: Measure blink peak values and compute a personalized threshold

Phase 3: Fine-tune the threshold for consistent and accurate detection

Runtime Signal Extraction:

During runtime, a rolling-median algorithm detects blinks based on deviations from the recent signal baseline, rather than a fixed amplitude cutoff.

i-Cognia: Enabling Independent Living

using Blink-Generated Brain Signals

i-Cognia is an assistive living system that enables independent control of communication and the home environment. It:

Detects blink-generated brain signals

Operates in real time

Allows users to navigate a hierarchical menu using eye blinks to:

Trigger emergency alerts

Connect with caregivers and family

Control home appliances (lights, fans, doors, etc.)

Access entertainment and cognitive exercises

Self-calibrates in under 30 seconds

Requires no clinician or specialized setup

Benefits of i-Cognia:

•Low-cost — built from commodity off-the-shelf components

•Personalize — freely adaptable for any patient, anywhere

•Non-invasive — no surgery, no implants, wearable at home

•User-Friendly — intuitive GUI for navigation of daily task

•Home-Ready — Tested for in-home setting in absence of specialized environments

* Does not need special lighting

* Does not need specialized posture

i-Cognia's System Oveview:

i-Cognia system is divided into four main components:

Input: This section captures user intent via an EEG module, detecting signals generated by eye movements or blinks.

I-Cognia Core: This acts as the main processing hub. It takes the raw EOG input and runs it through signal processing, feature extraction, and a command classifier. It also includes a GUI (Graphical User Interface) display that serves as the control panel.

Caregiver: A dedicated block connected to the GUI display, allowing a caregiver to monitor the system or receive alerts.

Output: The final stage where the classified commands are executed. The system triggers a relay module which acts as the physical switch to turn electrical home appliances on or off based on the user's eye blinks.

Evaluation and User-Testing:

To rigorously validate the technical viability and real-world usability of i-Cognia, I conducted an extensive evaluation comprising over 300 independent test runs across five different users. The primary objective was to measure raw signal processing performance and practical communication efficiency. For individuals with severe motor impairments, a communication device must be exceptionally reliable, so the system was tested to ensure it operates consistently outside of strictly controlled laboratory environments.

System performance for i‑Cognia is evaluated using several quantitative and subjective parameters that capture both technical accuracy in terms of signal level performance as well as navigation level performance and user experience. The overview of the performance parameters are:

Signal-Level Performance

Raw Blink Accuracy: Praw

Calibrated Blink Accuracy: PCalib

Navigation‑level performance

Navigation accuracy: PNav​

Navigation Throughput: NT

Navigation Information Transfer Rate: ITR

User experience

Task Load Index: NASA‑TLX

Novel Statistical Method Used for Capturing Eye Blinks Accurately:

i‑Cognia detetcts eye-blinks with high accuracy due to the use of proposed Median-Median Absolute Deviation (MAD) based algorithm. In particular, the median is used instead of the mean because EEG blink signals often contain large artifacts from movement, jaw activity, and noise.The proposed Median based MAD made calibration more stable across users and sessions, reduced false positives from noise spikes, and helped distinguish meaningful blink peaks from random artifacts, leading to a more reliable and responsive control signal.

i-Cognia Key Findings:

Extensive testing has shown that  i-Cognia is a viable, non-invasive assistive technology for individuals with severe motor impairments, such as ALS or locked-in syndrome. Key findings are given below:

i-Cognia achieved high Information Transfer Rate (ITR), demonstrating that users could communicate and execute commands efficiently.

Evaluation through the NASA Task Load Index (TLX) showed that i-Cognia is easy to use.

The prototype proved i-Cognia to be realistic, noisy environments by effectively maintaining accuracy.

The novel approach for Median MAD, based calibration clearly steers raw signals to the highly accurate signals.

The auto-calibration process successfully adapted to varying user baselines.

Triple-blink fail-safe drastically reduced false-positive selections caused by natural blinking or facial twitches.

i-Cognia's Median-MAD workflow

i-Cognia’s Median-MAD workflow improves assistive blink-control by reducing signal interference for better personalization and reliability. It further illustrates how this same strategy extends beyond i-Cognia to other biosignal systems like EEG, ECG, and wearables.

Stakeholders

A wide-range of stakeholders benefit from i-Cognia through greater independence, easier communication, reduced caregiving strain, improved care, and lower long-term support demands for many different people. These stakeholders include:

People with disabilities

Temporary patients

Caregivers

Families

Healthcare providers

Care homes

Insurance companies

Government

Market Scan

Existing solutions like sip-and-puff switches, surgical implants, and camera-based eye trackers are costly, invasive, and environment-dependent .

i-Cognia is more than 600% faster, over 99% less expensive, non-invasive, reliable, and requires minimal setup and training. Being fatigue-free, and widely accessible, i-Cognia is truly a breakthrough soultion.

Next Steps:

Boost Speed and Precision – Improve ITR, accuracy, and responsiveness for faster patient responses.

Age‑Informed Optimization – Study age‑related differences to fine‑tune system performance.

Universal Accessibility – Enhance usability across motor impairments (ALS, locked‑in, tetraplegia).

Partnerships for Scale – Scale through partnerships with government and disability organizations.

Path to Commercialization – Patent the core technology and launch a startup for real‑world impact.

Demos – Deliver an i-Cognia demo at the MaRS Discovery District, a Toronto innovation hub, and conduct a collaboration visit to the KITE Research Institute to explore validation and advancement in assistive technology and neurorehabilitation.

I acknowledge my family, especially grandparents, for their support, mentoring and encouragement. I also thank the Toronto Science Fair organization and my mentors for their invaluable support and guidance throughout this journey. I also sincerely thank OpenBCI for their generous grant.

Some images were generated using generative AI (Perplexity) based on design, logic, data and specifications provided by myself. All underlying data, visual design logic, patterns, and conclusions etc are my work.

Jonathan R. Wolpaw, J. R., Birbaumer, N., McFarland, D. J., Pfurtscheller, G., & Vaughan, T. M. (2002). Brain–computer interfaces for communication and control. Clinical Neurophysiology, 113(6), 767–791.

Gerwin Schalk, G., et al. (2004). BCI2000: A general-purpose brain-computer interface system. IEEE Transactions on Biomedical Engineering, 51(6), 1034–1043.

Dennis J. McFarland, D. J., & Wolpaw, J. R. (2011). Brain-computer interfaces for communication and control. Communications of the ACM, 54(5), 60–66.

Murat Akcakaya, M., Peters, B., Moghadamfalahi, M., Mooney, A., & Wolpaw, J. R. (2014). Noninvasive brain–computer interfaces for augmentative and alternative communication. IEEE Reviews in Biomedical Engineering, 7, 31–49.

Ahsan Khandoker, A. H., et al. (2017). EEG signal processing for eye blink detection: A review. Biomedical Signal Processing and Control, 33, 465–478.

Fabien Lotte, F., et al. (2018). A review of classification algorithms for EEG-based brain–computer interfaces. Journal of Neural Engineering, 15(3), 031005.

Robert Leeb, R., et al. (2013). Assistive technologies with brain–computer interfaces. Handbook of Clinical Neurology, 110, 437–445.

Luis Fernando Nicolas-Alonso, L. F., & Gomez-Gil, J. (2012). Brain computer interfaces, a review. Sensors, 12(2), 1211–1279.

World Health Organization. (2023). Assistive technology.

ALS Association. (2022). ALS statistics.

Statistics Canada. (2021). A portrait of family caregivers in Canada.

Teodor Mladenov, T. (2016). Disability and assistive technology: The importance of user-centered design. Disability & Society, 31(4), 566–579.

[AI/Large language model]. Used for grammatical improvements, text improvements, editing support, and image generation.