MOTIVATION LETTER Antimicrobial resistance kills nearly five million people each year, and sub-Saharan Africa carries the heaviest burden. When a patient in Lagos or Kano fails first-line treatment, clinicians rarely know whether the cause is a resistant mutation or poor drug quality. I built TOPOLOGIX to answer that question from sequence data alone. The system uses ESM-2 protein language model delta-embeddings combined with Morgan fingerprints and a Random Forest classifier to predict drug-resistance mutations, achieving an AUROC of 0.804 plus or minus 0.025 on the Platinum benchmark of 553 mutations. It covers 100 percent of mutations, where structure-based tools like mCSM-lig cover roughly 18 percent and score near 0.70. This is a tool designed for the fragmented diagnostic landscape of Nigerian hospitals, where sequencing infrastructure is scarce and reference laboratories are hours away. The Africa Healthcare Innovation Fellowship targets young African professionals building health-system solutions. I am a 29-year-old Nigerian pharmacist and computational researcher, currently enrolled in the M.Sc. Digital Health program at the Hasso Plattner Institute in Potsdam, Germany. My background spans clinical pharmacy, computational modeling, and software engineering. I have built four independent data ingestion and analysis pipelines using DuckDB, self-hosted local LLM serving with llama.cpp, and production systems on Linux VPS infrastructure. These are the practical toolkit required to take TOPOLOGIX from a benchmark-validated model to a field-tested prototype in a Nigerian clinical setting. The fellowship's 14-week field-based structure is precisely what TOPOLOGIX needs. The model has passed its computational validation gates. What it lacks is real-world clinical data from Nigerian patients, feedback from frontline pharmacists, and a deployment pathway that accounts for intermittent power, low bandwidth, and limited computational resources. I intend to use the fellowship period to design and run a small pilot study with two or three partner clinics, collecting bacterial isolate sequences and testing whether TOPOLOGIX's predictions hold outside benchmark conditions. I will also document the deployment requirements for a lightweight, offline-capable version of the tool. My broader research program informs this work. I have developed the Conjunctive Consolidation Threshold model for addiction neuroscience, a tripartite pharmacological framework with Bayesian MCMC calibration across 14 free parameters, and I have published negative results on topological methods for cardiotoxicity prediction, demonstrating that bipartite persistent homology does not beat descriptor baselines. These experiences taught me to design pre-registered validation pipelines and to report failures directly. TOPOLOGIX is the natural convergence of these skills: sequence representation learning, rigorous benchmarking, and a clear health-system problem. The fellowship's emphasis on collaboration across disciplines matches my working style. I have received endorsements from Kent Berridge at Michigan, Samuel Gershman at Harvard, Nathaniel Daw at Princeton, and Marcelo Mattar at NYU. I know how to work with clinicians, modelers, and software engineers. I am committed to the full May through July 2026 duration and to building a solution that stays in Nigeria after the fellowship ends. SHORT ESSAY: MOTIVATION AND ALIGNMENT WITH HEALTH-SYSTEM CHALLENGES Nigeria's health system is fragmented across public and private sectors, with diagnostic capacity concentrated in a few urban centers. Antimicrobial resistance compounds this fragmentation: clinicians prescribe empirically, resistance patterns go unmonitored, and data that could guide treatment decisions never reaches the point of care. TOPOLOGIX addresses this directly by predicting drug-resistance mutations from sequence data, removing the need for expensive structural analysis or extensive reference databases. The model's 0.804 AUROC on the Platinum benchmark and its 100 percent mutation coverage represent a practical improvement over structure-based alternatives that fail on 82 percent of mutations. My motivation is grounded in clinical experience. As a licensed pharmacist who has worked at Ramset Pharmacy and as National Product Manager at Synthcare, I have seen patients return with treatment failures and no clear path to adjust therapy. The tools to predict resistance exist in academic literature but not in clinical practice. The fellowship's focus on prototyping and field piloting is the missing link between my computational work and patient impact. I am proposing to build and test a deployment-ready tool during the fellowship period. The AHIF selection criteria emphasize demonstrated experience in healthcare and digital health, a strong interest in innovation, and the ability to collaborate across disciplines. My profile matches each criterion: a pharmacy degree from the University of Ibadan, an ongoing M.Sc. in Digital Health, four independent data pipelines, and a track record of collaboration with leading computational neuroscientists. I am committed to the full fellowship duration and to the practical, field-based work the programme requires. SHORT ESSAY: POTENTIAL FOR IMPACT AND COLLABORATION TOPOLOGIX's potential impact in Nigeria and across Africa rests on three concrete advantages. First, it requires only sequence data, which is increasingly accessible through portable sequencers and regional reference laboratories. Second, it covers the full mutation space, unlike structure-based tools that fail when crystal structures are unavailable. Third, it is computationally lightweight, designed to run on modest hardware. These properties make it deployable in settings where current tools simply do not function. The fellowship's collaborative structure will accelerate this work. I plan to partner with clinical microbiologists and pharmacists in two or three Nigerian hospitals to collect isolate sequences and validate predictions against phenotypic susceptibility testing. I will also engage with the fellowship cohort to learn from peers working on supply chain logistics, telemedicine, and community health worker tools. My experience building cross-disciplinary collaborations, evidenced by endorsements from researchers at Michigan, Harvard, Princeton, and NYU, prepares me for this work. I will measure impact in concrete terms: number of isolates processed, prediction accuracy against phenotypic results, and documented workflow changes in partner clinics. The fellowship's 14-week structure is sufficient to complete a small pilot and produce a deployment roadmap. The goal is a working tool that Nigerian clinicians can use. RESEARCH STATEMENT TOPOLOGIX is a sequence-based drug-resistance prediction system. It uses ESM-2 protein language model delta-embeddings to represent mutations, Morgan/ECFP fingerprints to represent drugs, and a Random Forest classifier to predict resistance. On the Platinum benchmark of 553 mutations, it achieves an AUROC of 0.804 plus or minus 0.025. On SKEMPI 2.0, it achieves 0.634. It outperforms structure-based baselines such as mCSM-lig, which scores approximately 0.70, while covering 100 percent of mutations compared to roughly 18 percent for structure-limited tools. The system emerged from a falsified hypothesis. My earlier work tested whether bipartite persistent homology of protein-ligand interfaces could predict hERG cardiotoxicity. A pre-registered, powered replication found that topological features do not beat a plain descriptor baseline, with an AUROC of 0.8426 versus 0.8782. A follow-up study applying the same topological constructs to drug-resistance prediction found almost no signal, with AUROCs of 0.425 and 0.485 on the Platinum benchmark. These negative results ruled out interface geometry as the driver of resistance and motivated the sequence-representation approach that became TOPOLOGIX. The current validation status is computational. The model has passed benchmark evaluations and outperforms existing tools on coverage and accuracy. What it lacks is real-world clinical validation. The fellowship period will address this gap through a pilot study in Nigerian clinical settings, collecting bacterial isolate sequences and comparing predictions against phenotypic susceptibility testing. My broader research program informs this work methodologically. I have developed the Conjunctive Consolidation Threshold model for addiction neuroscience, a tripartite ODE framework with Bayesian MCMC calibration across 14 free parameters, with all five pre-registered hypotheses confirmed. I have built neurocascade, a receptor-to-behavior brain-circuit simulation engine with 62 passing tests. I have extended Koopman operator methods with Mori-Zwanzig memory kernels for drug transport modeling, reporting a failed real-data gate directly rather than reframing it. These projects share a commitment to pre-registration, rigorous validation, and honest reporting of negative results. TOPOLOGIX is the project best matched to the Africa Healthcare Innovation Fellowship because it addresses a concrete health-system challenge, antimicrobial resistance, with a deployable computational tool. The fellowship's field-based structure and focus on prototyping align with the model's current stage: computationally validated, clinically unproven, and ready for real-world testing. EDITOR NOTES - Eligibility risk: The fellowship requires current residence in Africa. The applicant is enrolled at Hasso Plattner Institute in Potsdam, Germany for Winter Semester 2026/27. Verify whether the May to July 2026 fellowship period falls before the program start or whether remote participation is possible. This must be confirmed before submission. - The applicant's age, 29, and Nigerian citizenship satisfy the stated criteria, but the residence requirement is the primary risk. The motivation letter does not address this directly; the applicant should confirm their physical location during the fellowship period and adjust the letter if needed. - The pilot study described in the essays requires partner clinics. The applicant must insert specific clinic names or confirm that partnership discussions have begun. The current draft references "two or three partner clinics" without naming them, which is a gap a reviewer will notice. - The fellowship amount is unspecified. The application should not assume funding for the pilot study is included. The applicant should clarify whether the pilot requires separate funding and, if so, how it will be supported. - The applicant's employment as National Product Manager at Synthcare from March 2026 onward may conflict with the fellowship's full-time field-based structure. The applicant must confirm availability for the full May through July 2026 duration and note any leave arrangements. CHECKLIST - [ ] Confirm current residence in Africa for the May to July 2026 fellowship period - [ ] Confirm availability for full fellowship duration, including any employment conflicts with Synthcare - [ ] Identify and name specific partner clinics for the TOPOLOGIX pilot study - [ ] Verify whether the fellowship provides funding for pilot activities or whether separate funding is needed - [ ] Confirm the fellowship's application deadline and submission portal from the programme website - [ ] Prepare academic transcripts from University of Ibadan and Hasso Plattner Institute - [ ] Prepare CV listing research lines, publications, and technical skills - [ ] Secure two letters of recommendation, preferably from listed endorsers - [ ] Verify the fellowship's citizenship documentation requirements for Nigerian applicants - [ ] Submit the completed application before the deadline