You're pledging to donate if the project hits its minimum goal and gets approved. If not, your funds will be returned.
Brain-computer interfaces (BCIs) aim to restore communication for patients with ALS, a disease causing total immobility while the mind remains intact, yet these devices can fail unpredictably. Prior neurophysiology research shows measurable cortical hyperexcitability in ALS, present even before symptom onset in familial cases and worsening as the disease progresses. We track this using the EEG aperiodic exponent, extracted via spectral parameterization, and validated as a marker of excitation-inhibition (E:I) balance, to test whether it can predict when AI-assisted decoding can no longer restore communication. We hypothesize that if this marker shifts oppositely in a hyperexcitable condition (ALS) versus a hyperinhibited one like Parkinson's disease, consistent with external meta-analytic work (Norouzi et al., '25; Donoghue, '25), and scales gradually with sleep depth, BCI failure reflects a physical substrate constraint rather than a software limitation.
This work has been accepted for presentation at Models of Consciousness 7 (Copenhagen).
1. Re-analyze sessions within our six open datasets previously excluded by quality thresholds. Run a formal random-effects meta-analysis across the four independent Parkinson's datasets to test the bidirectional excitability hypothesis.
2. Present these re-analysis results at Models of Consciousness 7 in Copenhagen (October 2026) to secure expert feedback from the international consciousness-science community.
3. Submit the study protocol for ethics (IRB) review in parallel before enrolling new participants. Review timelines are expected to range from two weeks to three months.
4. Recruit and collect new EEG data from 6 ALS patients, 8 Parkinson's patients, and 8 sleep-study participants (22 total). Extract the aperiodic exponent marker and validate it using Mann-Whitney U tests and linear mixed-effects models.
5. Prepare the extended findings for peer-reviewed publication, building on two preprints already released from this research program.
Total ask of ~$4,300, itemized:
- Conference registration, travel, visa, and accommodation to present at MoC7 (5 nights): ~$2,800
- Human-subjects data collection funds, held for use once IRB approval clears: ~$2,100, including compensation for 6 ALS ($100 each), 8 Parkinson's ($75 each), and 8 sleep-study participants ($75 each), plus ~$300 for EEG consumables (caps, electrode gel) to extend the validation beyond the current cohort
- Cloud compute for re-analysis: $300, including GPU/API costs to run the breakeven-recall analysis on new sessions
- A0 color poster printing: $65 for poster presentation
I'm a self-taught neuroscientist (ORCID) and founder building non-invasive brain-computer interfaces that let late-stage ALS patients communicate while passively tracking cortical decline (Rayo). I've been trained through HarvardX, MIT, Neuromatch, and UIster's Computational Neuroscience programs. My prior work includes a mechanistic interpretability paper accepted to TMLR, and two preprints from this research program, namely Tran 2026a (aperiodic exponent in ALS BCI) and Tran 2026b (breakeven recall / LLM-assisted BCI decoding), currently under peer review. Research abstract also accepted to Models of Consciousness 7 (Copenhagen). Besides this main line of work, I'm also an active advocate and volunteer in multiple ALS and MND patient communities.
The biggest risk is the timeline: the Parkinson's meta-analysis might not resolve the bidirectional hypothesis cleanly, and I am committed to reporting that outcome as transparently as the null results from the individual datasets. On the logistics side, I am a Vietnamese citizen navigating Schengen visa timing against a registration deadline, but MoC7 has already issued a formal letter of invitation and I have a clear decision checkpoint if it does not come together in time.
For transparency, this project also has a small, parallel crowdfunding campaign on Experiment.com covering a similar scope; however, I am facing a verification roadblock because I am not a researcher based in the United States, United Kingdom, Canada, Australia, or Japan. This Manifund pitch is not asking for the same dollars twice. I'm happy to share current campaign status on request, and will update this page if the Experiment.com campaign is fully funded first.