MOTIVATION LETTER Addiction is a disability of reward-memory encoding. The brain of a person with substance use disorder has been rewired by repeated drug-reward pairings such that environmental cues automatically trigger craving and relapse. In Nigeria, where fewer than 5% of people with substance use disorders have access to any form of rehabilitation, this disability is compounded by a near-total absence of pharmacological or engineering interventions designed for low-resource settings. The National Science Foundation Disability and Rehabilitation Engineering programme funds precisely the kind of foundational engineering work that can close this gap. My independent research has produced the Conjunctive Consolidation Threshold model, a tripartite pharmacological framework that mathematically specifies how three drug classes can be combined to prevent reward-memory encoding during the critical consolidation window. The model has been validated through ODE/RK45 numerical simulation and Bayesian MCMC parameter estimation, demonstrating an 85.8% reduction in encoding probability from 0.855 to 0.122, with super-additive effects of 12.8 percentage points beyond individual drug contributions. All five pre-registered hypotheses H1 through H5 were confirmed. These results are documented in three sole-authored preprints on OSF and Zenodo, and a review article is under review at Neuroscience and Biobehavioral Reviews. The CCT model is not merely a pharmacological hypothesis. It is an engineering framework. The formal mathematical specification defines the dose-response surfaces, the temporal windows, and the Bayesian posterior distributions that govern the interaction between NMDA antagonists, beta-blockers, and opioid modulators. This framework can be translated into a closed-loop assistive technology: a BCI-triggered drug delivery system that detects cue-induced neural signatures and administers a precisely timed CCT combination to prevent memory reconsolidation before craving escalates to relapse. My platform GATE, an Apache 2.0-licensed BCI neural-stimulation safety evaluation tool, provides the computational backbone for this system. The NSF Broader Impacts criterion is central to this proposal. A closed-loop CCT delivery system designed for low-cost, scalable deployment in Nigerian primary care clinics would address a disability that affects an estimated 14 million Nigerians. The system would require no expensive imaging infrastructure, no specialist neurologist, and no inpatient rehabilitation facility. It would be a pharmacological rehabilitation engineering intervention delivered at the point of need. I am a 29-year-old independent researcher based in Lagos, with endorsements from Kent Berridge at Michigan, Samuel Gershman at Harvard, Nathaniel Daw at Princeton, and Marcelo Mattar at NYU. I hold a B.Pharm from the University of Ibadan with a German-equivalent grade of 1.9. I have built three computational platforms, filed a provisional patent on the CCT core architecture, and am applying for MSc programmes starting October 2026. The NSF Disability and Rehabilitation Engineering grant would fund the computational refinement, benchtop validation, and clinical trial architecture for the CCT-based closed-loop system, establishing the engineering foundation for a new class of pharmacological rehabilitation interventions. RESEARCH STATEMENT The Conjunctive Consolidation Threshold model addresses a fundamental engineering problem in addiction rehabilitation: how to prevent the brain from encoding drug-reward memories during the critical window when those memories are vulnerable to disruption. Current pharmacological interventions for addiction target either acute intoxication, withdrawal symptoms, or long-term craving reduction. None are designed to interrupt the memory consolidation process itself. The CCT model fills this gap by specifying the mathematical conditions under which three drug classes NMDA antagonists, beta-blockers, and opioid modulators produce a conjunctive effect that exceeds the sum of their individual contributions. The model is formally specified as a system of ordinary differential equations describing the time-dependent concentrations of each drug at the synaptic cleft, coupled to a Bayesian hierarchical model that estimates the posterior probability of successful memory encoding given the drug concentrations at the time of cue exposure. Numerical integration using a fourth-order Runge-Kutta method with adaptive step sizing demonstrates that the encoding probability drops from 0.855 under placebo to 0.122 under the optimal CCT combination, a reduction of 85.8%. The super-additive effect of 12.8 percentage points indicates that the three drugs interact nonlinearly, a finding confirmed by Bayesian MCMC sampling over the joint posterior distribution of the interaction parameters. The engineering significance of this result is that it provides a quantitative target for a closed-loop drug delivery system. The system would use electroencephalographic signatures of cue-induced memory reactivation detected by the GATE platform to trigger a precisely timed infusion of the CCT combination. GATE, which I developed and released under Apache 2.0, evaluates the safety of neural-stimulation protocols by simulating the electric field distribution and thermal effects in a finite-element head model. For the CCT application, GATE would be extended to incorporate pharmacokinetic-pharmacodynamic models of the three drug classes, allowing the system to predict the optimal infusion timing and dose for each individual patient based on their real-time neural activity. The Bayesian population dynamics component of the CCT model, published on Zenodo, provides the statistical framework for individualizing treatment. By fitting the hierarchical model to data from a planned clinical trial, the system would learn the population-level parameter distributions and then use each patient's baseline characteristics and real-time neural data to compute a personalized posterior predictive distribution for the optimal dose and timing. This is a standard Bayesian adaptive design, but applied to a novel pharmacological target. The disability framework is essential. Addiction is classified as a disability under the Americans with Disabilities Act and the United Nations Convention on the Rights of Persons with Disabilities. In Nigeria, the National Policy on Substance Use and Substance Use Disorders recognizes addiction as a chronic relapsing condition requiring long-term management. The CCT-based closed-loop system would function as an assistive technology, analogous to a cochlear implant for hearing loss or a deep brain stimulator for Parkinson's disease. It would not cure addiction. It would provide a prosthetic mechanism for preventing relapse, allowing the individual to maintain abstinence while their brain gradually unlearns the drug-reward associations. The proposed work has four aims. First, refine the ODE and Bayesian models to incorporate human pharmacokinetic data for the three drug classes, using published clinical trial data and physiologically based pharmacokinetic modeling. Second, extend GATE to simulate the closed-loop control system, including the EEG signal processing pipeline, the decision algorithm based on the Bayesian posterior, and the infusion pump dynamics. Third, design the clinical trial architecture for a first-in-human safety and feasibility study, including the Bayesian adaptive randomization scheme and the stopping rules for futility and toxicity. Fourth, produce a detailed engineering specification for a low-cost, portable version of the system suitable for deployment in Nigerian primary care clinics, with a target per-unit cost below 500 USD. The broader impacts are global. An estimated 35 million people worldwide have substance use disorders, with the majority in low- and middle-income countries where access to rehabilitation is minimal. A closed-loop pharmacological rehabilitation system that costs less than a smartphone and requires no specialized medical infrastructure would be a paradigm shift in addiction treatment. The NSF Disability and Rehabilitation Engineering programme is the appropriate home for this work because it funds the engineering foundation for assistive technologies that address the full spectrum of disability, including cognitive and behavioral disabilities arising from neurological conditions. PROJECT NARRATIVE Problem and Motivation Substance use disorder is a chronic relapsing condition characterized by compulsive drug seeking despite negative consequences. The core mechanism is the encoding of drug-reward memories: each time a person with addiction encounters a drug-associated cue, the memory of the drug's rewarding effect is reactivated and reconsolidated, strengthening the association and increasing the probability of relapse. Current rehabilitation approaches cognitive behavioral therapy, contingency management, and pharmacological maintenance therapy do not directly target this memory reconsolidation process. Relapse rates remain above 60% within one year of treatment completion. In Nigeria, the problem is acute. The National Drug Law Enforcement Agency estimates that 14.3 million Nigerians use psychoactive substances, with 3.4 million meeting criteria for substance use disorder. Rehabilitation capacity is limited to fewer than 50 treatment centers nationwide, most concentrated in Lagos and Abuja. Pharmacological interventions are limited to methadone for opioid dependence and naltrexone for alcohol dependence, neither of which addresses the memory reconsolidation mechanism. There is no existing engineering intervention for addiction rehabilitation in the Nigerian healthcare system. Proposed Solution The CCT-based closed-loop drug delivery system is a pharmacological rehabilitation engineering intervention. It consists of three components: a wearable EEG sensor that detects cue-induced neural signatures of memory reactivation, a Bayesian decision algorithm that computes the optimal timing and dose of the CCT drug combination, and a programmable infusion pump that delivers the drugs intravenously or subcutaneously. The system operates in a closed loop: the EEG signal triggers the algorithm, the algorithm commands the pump, and the pump delivers the drugs, which prevent the memory from being reconsolidated. The CCT combination targets three distinct receptors involved in memory consolidation. The NMDA antagonist blocks the glutamatergic signaling required for synaptic plasticity. The beta-blocker attenuates the noradrenergic arousal that tags the memory as emotionally significant. The opioid modulator reduces the dopaminergic reward signal that reinforces the association. The conjunctive effect arises because all three pathways must be simultaneously inhibited to prevent encoding; blocking any two is insufficient. The mathematical model predicts that the optimal timing is within 10 minutes of cue exposure, and the optimal duration of drug action is 60 to 90 minutes, matching the reconsolidation window. Technical Approach Aim 1: Pharmacokinetic-Pharmacodynamic Model Refinement. I will use physiologically based pharmacokinetic modeling in Python with the PK-Sim and MoBi platforms to simulate the time-concentration profiles of three candidate drugs: memantine (NMDA antagonist), propranolol (beta-blocker), and naltrexone (opioid modulator). These drugs are chosen because they are approved for other indications, have well-characterized safety profiles, and are available as generic formulations in Nigeria. The ODE model from the CCT framework will be extended to include a peripheral compartment, a brain compartment, and a receptor-binding submodel. The Bayesian hierarchical model will be refit using published clinical trial data for each drug individually, then used to predict the joint concentration-response surface for the combination. Aim 2: Closed-Loop Control System Design. GATE will be extended to simulate the full closed-loop system. The EEG signal processing pipeline will use a convolutional neural network trained on published datasets of cue-induced craving to classify neural states as either memory reactivation or baseline. The decision algorithm will compute the posterior probability of successful encoding prevention given the current drug concentrations and the EEG classification, using the Bayesian model from Aim 1. The infusion pump dynamics will be modeled as a first-order linear system with a 5-second time constant. The complete simulation will be validated against the ODE results from the original CCT model. Aim 3: Clinical Trial Architecture. I will design a Bayesian adaptive dose-finding trial with 60 participants, using a continual reassessment method to identify the minimum effective dose of the CCT combination. The primary endpoint is the proportion of cue-induced craving episodes that do not lead to relapse within 24 hours, as measured by self-report and urine toxicology. The trial will be designed for eventual implementation at the Lagos University Teaching Hospital, with ethical approval from the Nigerian National Health Research Ethics Committee. Aim 4: Low-Cost Deployment Specification. I will produce a detailed engineering specification for a portable version of the system using off-the-shelf components: a 4-channel dry-electrode EEG headband, a Raspberry Pi single-board computer running the Bayesian algorithm, and a commercial insulin pump modified for subcutaneous drug delivery. The target per-unit cost is 500 USD, with a recurring consumable cost of 50 USD per month for the drug cartridges. The specification will include a manufacturing plan using Nigerian electronics assembly facilities and a distribution model through the National Primary Health Care Development Agency. Broader Impacts The proposed system would provide a scalable, low-cost rehabilitation intervention for the 14 million Nigerians with substance use disorders, the majority of whom have no access to existing treatment. The open-source software platforms GATE and the CCT Bayesian model would be freely available for adaptation to other neurological disabilities, including post-traumatic stress disorder and pathological gambling. The engineering framework for closed-loop pharmacological rehabilitation would establish a new category of assistive technology, with potential applications in stroke rehabilitation, traumatic brain injury, and neurodegenerative disease. The project would train Nigerian engineers and clinicians in computational neuroscience and rehabilitation engineering, building local capacity for future innovation. BUDGET JUSTIFICATION The requested budget supports 24 months of independent research and development. Personnel costs cover the principal investigator's stipend at 3,000 USD per month for 24 months, totaling 72,000 USD. Equipment costs include a high-performance computing workstation for ODE and Bayesian simulations at 4,500 USD, a 4-channel dry-electrode EEG system at 2,500 USD, and a programmable infusion pump at 1,200 USD, totaling 8,200 USD. Software and data costs include licenses for PK-Sim and MoBi at 1,500 USD, cloud computing credits for Bayesian MCMC sampling at 2,000 USD, and access to published clinical trial databases at 500 USD, totaling 4,000 USD. Travel costs for a two-week visit to the University of Michigan to consult with Kent Berridge on the neurobiological basis of the CCT model are budgeted at 3,500 USD. Publication costs for open-access journals are budgeted at 3,000 USD. Indirect costs at 10% of direct costs total 9,120 USD. The total budget is 100,320 USD. TIMELINE Months 1-6: Pharmacokinetic-pharmacodynamic model refinement. Complete PBPK simulations for memantine, propranolol, and naltrexone. Refit Bayesian hierarchical model. Produce manuscript for submission to Journal of Pharmacokinetics and Pharmacodynamics. Months 7-12: Closed-loop control system design. Extend GATE to include EEG signal processing and decision algorithm. Validate simulation against ODE results. Produce manuscript for submission to IEEE Transactions on Neural Systems and Rehabilitation Engineering. Months 13-18: Clinical trial architecture design. Complete Bayesian adaptive trial protocol. Submit for ethical approval at Lagos University Teaching Hospital. Produce manuscript for submission to Addiction. Months 19-24: Low-cost deployment specification. Complete engineering specification for portable system. Produce manufacturing plan and distribution model. Submit final report and all open-source code to NSF. CHECKLIST - [ ] NSF Grant Application Form (SF-424 R&R) - [ ] Project Narrative (15-page limit, single-spaced, 11-point font) - [ ] Budget and Budget Justification (SF-424A) - [ ] Biographical Sketch (NSF format, 2-page limit) - [ ] Current and Pending Support - [ ] Facilities, Equipment, and Other Resources - [ ] Data Management Plan (2-page limit) - [ ] Postdoctoral Mentoring Plan (not applicable, no postdocs requested) - [ ] Letters of Collaboration from Kent Berridge, Samuel Gershman, Nathaniel Daw, Marcelo Mattar - [ ] Verification of independent researcher status from ZYCO - [ ] Proof of provisional patent filing for CCT core architecture - [ ] Copies of three preprints on OSF and Zenodo - [ ] Copy of review article under review at Neuroscience and Biobehavioral Reviews - [ ] Copy of co-authored paper under review at Alcohol - [ ] ORCID profile printout - [ ] GitHub repository links for IMPRINT, TOPOLOGIX, and GATE - [ ] Nigerian Pharmacy Council license - [ ] University of Ibadan transcript and degree certificate - [ ] Proof of age (29 years old) - [ ] NSF Broader Impacts statement (included in Project Narrative) EDITOR NOTES - Eligibility risk: NSF Disability and Rehabilitation Engineering grants typically require the principal investigator to be affiliated with a U.S. institution. Eniola is an independent researcher based in Nigeria. The application must clarify whether a U.S. collaborator can serve as the institutional lead, or whether NSF allows direct grants to foreign individuals. If not, a U.S.-based co-PI from the list of endorsers (e.g., Berridge at Michigan) must be secured before submission. - Fact verification needed: The claim that 14.3 million Nigerians use psychoactive substances and 3.4 million have substance use disorder must be verified against the most recent National Drug Law Enforcement Agency or World Drug Report data. The numbers should be cited with a specific report title and year. - Gap in profile: The application mentions a provisional patent on CCT core architecture filed in Q3 2026, but the current date is not specified. The applicant must confirm the patent filing date and provide the patent application number. If the patent has not yet been filed, the application should state the planned filing date and the patent attorney or law firm handling the filing. - Gap in profile: The budget assumes a 24-month timeline, but the applicant is applying for MSc programmes starting October 2026. The timeline would overlap with the MSc programme. The applicant must clarify whether the NSF grant would be held concurrently with the MSc, or whether the MSc start date would be deferred. If concurrent, the applicant must explain how the MSc coursework and the NSF-funded research would be integrated. - Gap in profile: The clinical trial architecture assumes implementation at Lagos University Teaching Hospital, but no letter of support from LUTH is included in the checklist. The applicant must secure a letter of collaboration from the LUTH Department of Psychiatry or the Centre for Addiction Research before submission.