Nyx Dynamics LLC
Nyx Dynamics LLC develops clinically grounded computational methods for healthcare AI, biomedical evidence review, and medical decision support.
Research and methods are published openly. Commercial product lines operate separately, on their own sites.
Our Origin Story
"A reinvention born from profound challenge and growth — a reclaiming of voice, skills, and purpose."
— A.C. Demidont, Founder
Clinical medicine and computation in a single practice
Nyx Dynamics applies computational epidemiology to problems of structural harm — how systems, biological and institutional, produce irreversible damage through timing-dependent processes. The work pairs two decades of clinical infectious-diseases experience with the computational methods needed to build and validate the models directly. Research output is open-access, with code and data deposited publicly.
Two decades of board-certified infectious diseases practice, including Chief Medical Officer at Anchor Health Initiative and Principal Medical Scientist at Gilead Sciences.
Bayesian and stochastic modeling, simulation at population scale, and reference-structure analysis of the biomedical literature — built, validated, and shipped as working tools.
Eight active research programs spanning HIV prevention implementation science, algorithmic bias epidemiology, mathematical prevention theory, quantum biology of HIV neuropathogenesis, Bayesian neurometabolic modeling, and reproducibility policy — two peer-reviewed publications and six posted preprints, with code and data deposited openly. Full portfolio at /research.
MIT Sloan Executive Education in AI for Healthcare; the MITx Computational Thinking with Python series (6.00.1x, 6.00.2x); and graduate coursework from the MITx Statistics and Data Science MicroMasters (6.431x, IDS.S24x).
Founded and led by A.C. Demidont, DO — Founder & Principal Investigator.
An algorithm isn't good or bad per se – it is just a tool. Who is using it, and for which patients? We go beyond accuracy metrics.
Rigorous validation exposes uncertainty rather than eliminating it. Computationally ready ≠ clinically proven.
There cannot be a universal checklist – clinical context is too important. Human judgment will always be essential.
When is a healthcare algorithm ready for deployment?
GitHub · Zenodo DOI · Published in Viruses 2026 as Supplementary S3
Is sufficient high-quality data available?
Does precision create false confidence?
Can clinicians trust the parameters?
Does transparency enable oversight?
Do averages mask disparities?
Should implementation proceed?
"An algorithm that appears to have all answers is dangerous. One that names its limitations invites trust."
— The Readiness Paradox
Jun 6, 2026
Irreversible proviral integration defines a finite, route-specific window during which PEP can succeed. For parenteral (injection) exposure, this window compresses ~3× versus mucosal exposure — to 16–28 hours. Structural access delays for PWID systematically eliminate it, bounding population-level PEP efficacy below 10%. A structural failure, not a pharmacological one.
Jan 2026
Computational validation of a clinical decision support algorithm for LAI-PrEP bridge period navigation, validated at UNAIDS global target scale.
Jan 2025
Nyx Dynamics founder completes MIT Sloan Executive Education program on AI implementation in clinical settings.
Dec 2024
Our six-question framework for evaluating clinical AI readiness now available for healthcare organizations. GitHub · Zenodo
Oct 2024
Completed equity audit of PrEP eligibility algorithms, identifying disparities in underserved populations.