Closed-loop Neuroelectronic Interfaces: In Vitro to Silicon to Clinical Translation
Bibliographic record
Abstract
Neuroelectronic interfaces enable interaction between biological neural systems and artificial computation. This technology can be applied with the objectives of treating neurological disorders, augmenting biological functionality, and understanding basic neurophysiology. This dissertation outlines the development of devices that actively interpret and control electrical activity in neural tissue. More specifically, interpretation is performed using machine learning to classify neural states which have been learned from pre-recorded examples, and control is achieved by applying electrical waveforms that induce physiological activity. The target application is epilepsy, a disorder that causes seizures in 65 million people worldwide. The underlying brain states in this disorder are distinct and thus classifiable. Creating a link between classification and responsive electrical stimulation to provide meaningful disease treatment is known as closed-loop neuromodulation. This dissertation explores closed-loop neuromodulation in a lab environment (in vitro), using silicon integrated circuits to enable implantable devices (in vivo), and the clinical translation process. Microelectrode arrays (MEAs) enable the in vitro research of disease-relevant neural circuitry and the testing of device-tissue interaction dynamics using brain slices. The OpenMEA system was developed to overcome limitations that hinder both fundamental neuromodulation research and medical device development. A <1ms closed-loop latency is achieved by using a tightly-coupled FPGA, and a novel microfluidics system prevents perfusion-associated microscopy distortion and improves electrode-tissue adherence. Translating discoveries made in vitro to medical devices used in vivo requires the creation of application-specific integrated circuits (ASICs) to support life-long operation with a limited battery capacity. The BrainForest closed-loop system-on-chip is introduced, featuring innovations in low-power neuromorphic computer architecture. BrainForest achieves a seizure sensitivity of >99% and a false detection rate of 0.84 per hour with a power consumption of 118 uW. This enables multi-year personalized neuromodulation using existing battery technology. The clinical translation of neuromodulation ASICs is explored through a pilot study conducted at the Toronto Western Hospital under research ethics board approval. The final chapter outlines the challenges of deployment in a clinical setting, and the solutions developed to convert large volumes of patient data into efficient machine learning models that can run on a chronically implanted device.
Fetched live from OpenAlex and de-inverted. Abstracts are not stored in this database: the inverted indexes are 8.6 GB of the frame’s 9.3 GB of text, and the host has 13 GB free.
How this classification was reachedexpand
Full frame machine prediction
Teacher imitationNot calibrated prevalence, not ground truth. Human validation pending. The Gemma side is a direct model label for every work in the frame, read from the title-only record. The Codex side is a classifier learned from the 10,348 direct Codex labels and calibrated to design-weighted sample rates; fields without enough sample support carry no Codex call. Candidate is the union of the two sides; consensus is their intersection. These outputs are machine_predicted_unvalidated and are not human labels.
Distilled classifier scores by category (both heads)
| Category | Codex | Gemma |
|---|---|---|
| Metaresearch | 0.001 | 0.002 |
| Meta-epidemiology (narrow) | 0.000 | 0.000 |
| Meta-epidemiology (broad) | 0.000 | 0.000 |
| Bibliometrics | 0.000 | 0.000 |
| Science and technology studies | 0.000 | 0.001 |
| Scholarly communication | 0.002 | 0.001 |
| Open science | 0.001 | 0.001 |
| Research integrity | 0.001 | 0.001 |
| Insufficient payload (model declined to judge) | 0.003 | 0.001 |
Machine scores (provisional)
The two teacher heads of the student model, read on this work. A score orders the frame for review; it never asserts a category, and the validation status ships verbatim with every row.
Baseline scores from an immature model (maturity gate not passed, 7 training rounds). Scores rank; they never assert a category.
score_only:v0-immature-baseline · verbatim from the scoring run: score_only means the number may rank works, and no category label ships from itClassification
machine, unvalidatedMachine predicted; a candidate call from one source (direct Gemma or distilled Codex), not a consensus.
How this classification was reached, model by model and score by score, is at the end of the page under "How this classification was reached".