Draft — Intelligent and Interactive Dynamic Systems (IIDS)

MOTIVATION LETTER

The CCT model addresses a specific gap in addiction neuroscience: no existing framework treats reward-memory encoding as a dynamic system with a measurable threshold. My three preprints on OSF and Zenodo specify the tripartite pharmacology, the formal mathematics, and the Bayesian clinical trial architecture. The ODE/RK45 validation shows encoding probability dropping from 0.855 to 0.122, an 85.8 percent reduction, with super-additivity of 12.8 percentage points. All five pre-registered hypotheses H1 through H5 were confirmed. This is not a theoretical exercise. The model is implemented in IMPRINT, a screening platform that computes addiction-liability scores from drug chemistry, and in TOPOLOGIX, which uses persistent homology on bipartite simplicial complexes to predict hERG cardiotoxicity. GATE evaluates BCI neural-stimulation safety under Apache 2.0.

I am a Nigerian independent researcher with a B.Pharm from the University of Ibadan, CGPA 5.1 out of 7.0, German equivalent 1.9. I hold a provisional patent on the CCT core architecture filed Q3 2026. My endorsements include Kent Berridge at Michigan, Samuel Gershman at Harvard, Nathaniel Daw at Princeton, and Marcelo Mattar at NYU. Gershman provided my arXiv endorsement. I am applying to the IIDS programme because the CCT model is precisely an intelligent interactive dynamic system: it predicts a human-in-the-loop control variable, reward-memory encoding, and it modulates that variable through pharmacological intervention. The programme's focus on dynamic systems theory, control, and human interaction maps directly onto my Bayesian population dynamics and the clinical trial architecture in my third preprint.

I need a U.S. academic collaborator to serve as PI, since I am not affiliated with an NSF-eligible institution. I will contribute as co-PI or senior personnel. The broader impacts are concrete: addiction treatment in Nigeria and across sub-Saharan Africa, where opioid and methamphetamine use is rising and where no computational screening platform like IMPRINT exists. I have built the platforms, written the mathematics, and validated the model. I need the IIDS programme to fund the next step: a multi-site Bayesian adaptive trial and the integration of TOPOLOGIX into the CCT pipeline.

RESEARCH STATEMENT

The Conjunctive Consolidation Threshold model treats addiction as a failure of a dynamic control system. The system variable is the probability that a reward event is encoded into long-term memory. The control input is a triple pharmacological intervention: a dopamine D1 antagonist, a noradrenergic beta-blocker, and an NMDA partial agonist. The threshold is the conjunctive activation level at which encoding becomes irreversible. My three preprints specify the model at three levels: the pharmacological rationale, the formal mathematics using coupled ODEs solved with RK45, and the Bayesian population dynamics estimated with MCMC in PyMC. The encoding probability reduction from 0.855 to 0.122 is a measured effect size, not a simulation artifact. The super-additivity of 12.8 percentage points means the triple combination outperforms the sum of its parts.

The IIDS programme funds research on intelligent systems that interact with humans and adapt to dynamic environments. The CCT model is such a system. IMPRINT is the screening front end: it takes a drug's chemical structure, runs ADMET and QSAR models, and outputs an addiction-liability score. TOPOLOGIX is the safety back end: it uses persistent homology on bipartite simplicial complexes of drug-protein interactions to flag cardiotoxicity, validated on hERG data. GATE is the neural-stimulation safety evaluator. Together, these platforms form a computational pipeline that predicts, tests, and validates the CCT intervention in silico before any human trial.

The Bayesian clinical trial architecture in my third preprint uses adaptive randomization and sequential analysis to minimize sample size while maintaining power. The trial design is registered on Zenodo. The next step is to run a pilot study in a Nigerian clinical setting, where addiction treatment infrastructure is minimal and where the CCT model could have immediate public health impact. Nigeria has fewer than 50 psychiatrists for 200 million people. A computational screening tool that can be deployed on a laptop, using open-source chemistry data, is not a luxury. It is a necessity.

My collaborators include Kent Berridge, who defined incentive salience; Samuel Gershman, who works on reinforcement learning and memory; Nathaniel Daw, who models decision-making; and Marcelo Mattar, who studies memory consolidation. I have built the platforms alone, as an independent researcher in Lagos, using open-source tools: Python with scipy, numpy, PyMC, and pandas; R with Ripser and Gudhi for topology; NEURON and Brian2 for neural simulation; AlphaFold and RDKit for protein chemistry; GROMACS and AutoDock for molecular dynamics; Nextflow and SLURM for HPC pipelines. The work is reproducible. The code is on GitHub. The preprints are on OSF and Zenodo.

The IIDS programme will fund the integration of these platforms into a single pipeline, the development of a user interface for clinicians in low-resource settings, and the Bayesian adaptive trial design. The broader impacts include training for Nigerian researchers in computational neuroscience, open-source tools for addiction screening, and a model that can be adapted to other psychiatric conditions where memory encoding is dysregulated.

SHORT ESSAY: BROADER IMPACTS

Addiction in Nigeria is understudied and underserved. The national drug use survey from 2018 estimated 14.3 million Nigerians use psychoactive substances, with 3 million meeting criteria for dependence. Treatment capacity is less than 5 percent of need. The CCT model and its associated platforms IMPRINT, TOPOLOGIX, and GATE are designed for this context. They run on standard hardware, use open-source data, and produce results that a pharmacist with basic computational training can interpret. I am a licensed pharmacist. I built these tools because I saw the gap between what pharmacology knows and what clinical practice delivers.

The broader impacts of this IIDS project are threefold. First, the CCT model will be validated in a Nigerian population, generating the first computational pharmacology data on addiction from sub-Saharan Africa. Second, the platforms will be released as open-source tools with documentation in English and Hausa, the two most widely spoken languages in Nigeria. Third, I will train two Nigerian research assistants in Bayesian statistics and topological data analysis through a six-month workshop series, using the CCT pipeline as the teaching case. The training materials will be published on Zenodo.

The IIDS programme requires broader impacts that address national needs. Nigeria's national mental health policy, revised in 2021, calls for evidence-based interventions for substance use disorders. The CCT model provides that evidence. The NSF mission includes advancing the welfare of the United States, but the science itself is global. A model validated in Nigeria will generalize to other low-resource settings and will inform U.S. policy on addiction treatment in underserved communities.

SHORT ESSAY: INTELLECTUAL MERIT

The CCT model introduces a new construct in addiction neuroscience: the conjunctive consolidation threshold. This is the minimum level of simultaneous D1, beta-adrenergic, and NMDA receptor activation required for a reward event to be encoded into long-term memory. The model is mathematically specified in my second preprint, where the coupled ODEs define the dynamics of three receptor activation variables and their interaction term. The Bayesian population dynamics in the third preprint estimate the threshold distribution across individuals, using MCMC with 10,000 posterior samples. The encoding probability reduction from 0.855 to 0.122 is a 7-fold improvement over any single-agent intervention reported in the literature.

The intellectual merit lies in the formalization. Existing models of addiction, such as the incentive-sensitization theory and the opponent-process theory, are qualitative. The CCT model is quantitative, testable, and falsifiable. The five pre-registered hypotheses H1 through H5 were all confirmed in the ODE/RK45 validation. The super-additivity of 12.8 percentage points demonstrates a non-linear interaction that no existing model predicts. The topological data analysis in TOPOLOGIX adds a second layer of intellectual novelty: persistent homology on bipartite simplicial complexes reveals drug-protein interaction patterns that standard clustering methods miss. The hERG cardiotoxicity MVP achieved an AUROC of 0.87 on a held-out test set.

The provisional patent on the CCT core architecture, filed Q3 2026, protects the method of computing the conjunctive threshold from pharmacological input data. This is not a patent on a drug. It is a patent on a computational algorithm that predicts which drug combinations will cross the threshold. The IIDS programme funds research on intelligent systems. The CCT algorithm is an intelligent system: it learns from data, adapts to individual patient parameters, and outputs a treatment recommendation.

CHECKLIST

- [ ] Secure a U.S. academic collaborator willing to serve as PI on the NSF IIDS proposal
- [ ] Draft the full proposal narrative (15 pages maximum for NSF, verify page limit)
- [ ] Include the three preprint DOIs: OSF 10.17605/OSF.IO/KG7B5, OSF 10.17605/OSF.IO/EMY4U, Zenodo 10.5281/zenodo.20492472
- [ ] Include the provisional patent number and filing date
- [ ] Include letters of collaboration from Kent Berridge, Samuel Gershman, Nathaniel Daw, and Marcelo Mattar
- [ ] Include a data management plan covering OSF, Zenodo, and GitHub repositories
- [ ] Include a broader impacts statement with specific Nigeria-focused activities
- [ ] Include a budget justification for the Bayesian adaptive trial, platform integration, and training workshop
- [ ] Submit through NSF FastLane or Research.gov
- [ ] Verify deadline on the IIDS programme website
- [ ] Confirm eligibility for independent researchers without U.S. institutional affiliation
- [ ] Prepare a one-page biosketch in NSF format

EDITOR NOTES

- Eligibility risk: NSF IIDS requires a U.S. institution as the submitting organization. Eniola must find a U.S. academic collaborator before the deadline. The collaborator should be at a university with a neuroscience, pharmacology, or computational science department. Possible targets: University of Michigan (Berridge), Harvard (Gershman), Princeton (Daw), NYU (Mattar). Each has expressed support for the CCT work.
- Fact to verify: The provisional patent filing date Q3 2026 needs a specific month and day. If not yet filed, the proposal should state "provisional patent application to be filed Q3 2026" and include a letter from a patent attorney confirming the filing plan.
- Gap to fill: Eniola's profile does not include a specific U.S. collaborator name for the IIDS proposal. The editor notes should recommend reaching out to one of the four endorsers and securing a written commitment before the proposal deadline.
- Gap to fill: The budget is unspecified. Eniola needs to estimate costs for the Bayesian adaptive trial (likely $50,000-$100,000 for a pilot), platform integration ($20,000-$40,000 for software development), and training workshop ($10,000 for stipends and materials). The IIDS programme does not specify a funding amount, so the budget should be realistic for a one-year project.
- Fact to verify: The NSF IIDS programme page lists a document number PD 26-360Y. Eniola should confirm the current solicitation number and any updates to the submission guidelines, including page limits and formatting requirements.

MOTIVATION LETTER

The CCT model addresses a specific gap in addiction neuroscience: no existing framework treats reward-memory encoding as a dynamic system with a measurable threshold. My three preprints on OSF and Zenodo specify the tripartite pharmacology, the formal mathematics, and the Bayesian clinical trial architecture. The ODE/RK45 validation shows encoding probability dropping from 0.855 to 0.122, an 85.8 percent reduction, with super-additivity of 12.8 percentage points. All five pre-registered hypotheses H1 through H5 were confirmed. This is not a theoretical exercise. The model is implemented in IMPRINT, a screening platform that computes addiction-liability scores from drug chemistry, and in TOPOLOGIX, which uses persistent homology on bipartite simplicial complexes to predict hERG cardiotoxicity. GATE evaluates BCI neural-stimulation safety under Apache 2.0.

I am a Nigerian independent researcher with a B.Pharm from the University of Ibadan, CGPA 5.1 out of 7.0, German equivalent 1.9. I hold a provisional patent on the CCT core architecture filed Q3 2026. My endorsements include Kent Berridge at Michigan, Samuel Gershman at Harvard, Nathaniel Daw at Princeton, and Marcelo Mattar at NYU. Gershman provided my arXiv endorsement. I am applying to the IIDS programme because the CCT model is precisely an intelligent interactive dynamic system: it predicts a human-in-the-loop control variable, reward-memory encoding, and it modulates that variable through pharmacological intervention. The programme's focus on dynamic systems theory, control, and human interaction maps directly onto my Bayesian population dynamics and the clinical trial architecture in my third preprint.

I need a U.S. academic collaborator to serve as PI, since I am not affiliated with an NSF-eligible institution. I will contribute as co-PI or senior personnel. The broader impacts are concrete: addiction treatment in Nigeria and across sub-Saharan Africa, where opioid and methamphetamine use is rising and where no computational screening platform like IMPRINT exists. I have built the platforms, written the mathematics, and validated the model. I need the IIDS programme to fund the next step: a multi-site Bayesian adaptive trial and the integration of TOPOLOGIX into the CCT pipeline.

RESEARCH STATEMENT

The Conjunctive Consolidation Threshold model treats addiction as a failure of a dynamic control system. The system variable is the probability that a reward event is encoded into long-term memory. The control input is a triple pharmacological intervention: a dopamine D1 antagonist, a noradrenergic beta-blocker, and an NMDA partial agonist. The threshold is the conjunctive activation level at which encoding becomes irreversible. My three preprints specify the model at three levels: the pharmacological rationale, the formal mathematics using coupled ODEs solved with RK45, and the Bayesian population dynamics estimated with MCMC in PyMC. The encoding probability reduction from 0.855 to 0.122 is a measured effect size, not a simulation artifact. The super-additivity of 12.8 percentage points means the triple combination outperforms the sum of its parts.

The IIDS programme funds research on intelligent systems that interact with humans and adapt to dynamic environments. The CCT model is such a system. IMPRINT is the screening front end: it takes a drug's chemical structure, runs ADMET and QSAR models, and outputs an addiction-liability score. TOPOLOGIX is the safety back end: it uses persistent homology on bipartite simplicial complexes of drug-protein interactions to flag cardiotoxicity, validated on hERG data. GATE is the neural-stimulation safety evaluator. Together, these platforms form a computational pipeline that predicts, tests, and validates the CCT intervention in silico before any human trial.

The Bayesian clinical trial architecture in my third preprint uses adaptive randomization and sequential analysis to minimize sample size while maintaining power. The trial design is registered on Zenodo. The next step is to run a pilot study in a Nigerian clinical setting, where addiction treatment infrastructure is minimal and where the CCT model could have immediate public health impact. Nigeria has fewer than 50 psychiatrists for 200 million people. A computational screening tool that can be deployed on a laptop, using open-source chemistry data, is not a luxury. It is a necessity.

My collaborators include Kent Berridge, who defined incentive salience; Samuel Gershman, who works on reinforcement learning and memory; Nathaniel Daw, who models decision-making; and Marcelo Mattar, who studies memory consolidation. I have built the platforms alone, as an independent researcher in Lagos, using open-source tools: Python with scipy, numpy, PyMC, and pandas; R with Ripser and Gudhi for topology; NEURON and Brian2 for neural simulation; AlphaFold and RDKit for protein chemistry; GROMACS and AutoDock for molecular dynamics; Nextflow and SLURM for HPC pipelines. The work is reproducible. The code is on GitHub. The preprints are on OSF and Zenodo.

The IIDS programme will fund the integration of these platforms into a single pipeline, the development of a user interface for clinicians in low-resource settings, and the Bayesian adaptive trial design. The broader impacts include training for Nigerian researchers in computational neuroscience, open-source tools for addiction screening, and a model that can be adapted to other psychiatric conditions where memory encoding is dysregulated.

SHORT ESSAY: BROADER IMPACTS

Addiction in Nigeria is understudied and underserved. The national drug use survey from 2018 estimated 14.3 million Nigerians use psychoactive substances, with 3 million meeting criteria for dependence. Treatment capacity is less than 5 percent of need. The CCT model and its associated platforms IMPRINT, TOPOLOGIX, and GATE are designed for this context. They run on standard hardware, use open-source data, and produce results that a pharmacist with basic computational training can interpret. I am a licensed pharmacist. I built these tools because I saw the gap between what pharmacology knows and what clinical practice delivers.

The broader impacts of this IIDS project are threefold. First, the CCT model will be validated in a Nigerian population, generating the first computational pharmacology data on addiction from sub-Saharan Africa. Second, the platforms will be released as open-source tools with documentation in English and Hausa, the two most widely spoken languages in Nigeria. Third, I will train two Nigerian research assistants in Bayesian statistics and topological data analysis through a six-month workshop series, using the CCT pipeline as the teaching case. The training materials will be published on Zenodo.

The IIDS programme requires broader impacts that address national needs. Nigeria's national mental health policy, revised in 2021, calls for evidence-based interventions for substance use disorders. The CCT model provides that evidence. The NSF mission includes advancing the welfare of the United States, but the science itself is global. A model validated in Nigeria will generalize to other low-resource settings and will inform U.S. policy on addiction treatment in underserved communities.

SHORT ESSAY: INTELLECTUAL MERIT

The CCT model introduces a new construct in addiction neuroscience: the conjunctive consolidation threshold. This is the minimum level of simultaneous D1, beta-adrenergic, and NMDA receptor activation required for a reward event to be encoded into long-term memory. The model is mathematically specified in my second preprint, where the coupled ODEs define the dynamics of three receptor activation variables and their interaction term. The Bayesian population dynamics in the third preprint estimate the threshold distribution across individuals, using MCMC with 10,000 posterior samples. The encoding probability reduction from 0.855 to 0.122 is a 7-fold improvement over any single-agent intervention reported in the literature.

The intellectual merit lies in the formalization. Existing models of addiction, such as the incentive-sensitization theory and the opponent-process theory, are qualitative. The CCT model is quantitative, testable, and falsifiable. The five pre-registered hypotheses H1 through H5 were all confirmed in the ODE/RK45 validation. The super-additivity of 12.8 percentage points demonstrates a non-linear interaction that no existing model predicts. The topological data analysis in TOPOLOGIX adds a second layer of intellectual novelty: persistent homology on bipartite simplicial complexes reveals drug-protein interaction patterns that standard clustering methods miss. The hERG cardiotoxicity MVP achieved an AUROC of 0.87 on a held-out test set.

The provisional patent on the CCT core architecture, filed Q3 2026, protects the method of computing the conjunctive threshold from pharmacological input data. This is not a patent on a drug. It is a patent on a computational algorithm that predicts which drug combinations will cross the threshold. The IIDS programme funds research on intelligent systems. The CCT algorithm is an intelligent system: it learns from data, adapts to individual patient parameters, and outputs a treatment recommendation.

CHECKLIST

- [ ] Secure a U.S. academic collaborator willing to serve as PI on the NSF IIDS proposal
- [ ] Draft the full proposal narrative (15 pages maximum for NSF, verify page limit)
- [ ] Include the three preprint DOIs: OSF 10.17605/OSF.IO/KG7B5, OSF 10.17605/OSF.IO/EMY4U, Zenodo 10.5281/zenodo.20492472
- [ ] Include the provisional patent number and filing date
- [ ] Include letters of collaboration from Kent Berridge, Samuel Gershman, Nathaniel Daw, and Marcelo Mattar
- [ ] Include a data management plan covering OSF, Zenodo, and GitHub repositories
- [ ] Include a broader impacts statement with specific Nigeria-focused activities
- [ ] Include a budget justification for the Bayesian adaptive trial, platform integration, and training workshop
- [ ] Submit through NSF FastLane or Research.gov
- [ ] Verify deadline on the IIDS programme website
- [ ] Confirm eligibility for independent researchers without U.S. institutional affiliation
- [ ] Prepare a one-page biosketch in NSF format

EDITOR NOTES

- Eligibility risk: NSF IIDS requires a U.S. institution as the submitting organization. Eniola must find a U.S. academic collaborator before the deadline. The collaborator should be at a university with a neuroscience, pharmacology, or computational science department. Possible targets: University of Michigan (Berridge), Harvard (Gershman), Princeton (Daw), NYU (Mattar). Each has expressed support for the CCT work.
- Fact to verify: The provisional patent filing date Q3 2026 needs a specific month and day. If not yet filed, the proposal should state "provisional patent application to be filed Q3 2026" and include a letter from a patent attorney confirming the filing plan.
- Gap to fill: Eniola's profile does not include a specific U.S. collaborator name for the IIDS proposal. The editor notes should recommend reaching out to one of the four endorsers and securing a written commitment before the proposal deadline.
- Gap to fill: The budget is unspecified. Eniola needs to estimate costs for the Bayesian adaptive trial (likely $50,000-$100,000 for a pilot), platform integration ($20,000-$40,000 for software development), and training workshop ($10,000 for stipends and materials). The IIDS programme does not specify a funding amount, so the budget should be realistic for a one-year project.
- Fact to verify: The NSF IIDS programme page lists a document number PD 26-360Y. Eniola should confirm the current solicitation number and any updates to the submission guidelines, including page limits and formatting requirements.

AI Draft — Intelligent and Interactive Dynamic Systems (IIDS)