Preventive medicine & occupational medicine physician. Epidemiologist. Building the science that clinical AI deployment has been missing since the beginning.
Over 700 AI systems are active in clinical medicine. Almost none are monitored after deployment. That is not a technology problem. It is a surveillance problem.
The Epidemiology of Algorithms is the discipline I am building to close that gap — the science of post-deployment surveillance for AI systems in clinical care. It applies the methods of epidemiology to a new class of exposure: the algorithm itself.
"What do AI systems and influenza viruses have in common? They both shift and drift — and we have built a global surveillance system for one of them."
When an AI system is deployed in a clinical environment, it becomes a population-level intervention. Its effects are distributed across every patient touched by every clinician who uses it. A sepsis algorithm that starts missing cases does not fail one patient — it fails a population, invisibly, until someone builds the infrastructure to see it.
We built that infrastructure for influenza after 1918. The WHO's Global Influenza Surveillance and Response System now coordinates sentinel sites across 114 countries, sequences circulating strains continuously, and detects drift in real time. The entire architecture exists because we recognized that a pathogen that shifts and drifts requires ongoing population-level monitoring.
Clinical AI shifts and drifts. No equivalent surveillance infrastructure exists. This newsletter — and the discipline it is building — exists to change that.
I attended the Johns Hopkins Bloomberg School of Public Health, where I earned a Master of Health Science in epidemiology and infectious diseases. I then attended medical school at Howard University College of Medicine, followed by a residency in Ophthalmology. After a few years of practice, I missed epidemiology and went back to complete a residency in preventive medicine at the University of Maryland Medical School. During that residency, I obtained a Master of Science in Epidemiology and Biostatistics.
I have practiced for the last 20+ years as a preventive and occupational medicine physician, specialties that have always been concerned with population-level exposures and their effects on health — the same framework I am now applying to algorithms. I am dual board-certified in preventive medicine and lifestyle medicine.
Before Hopkins, I worked as a laboratory technician. I used HeLa cells in tissue culture routinely — without knowing their origins. It was only at Hopkins that I learned the full story of Henrietta Lacks and understood what extraction without consent looks like from the inside: ordinary, unremarkable, invisible until someone names it. That experience shapes how I think about health data, algorithmic training sets, and the ethical architecture of AI deployment.
The Epidemiology of Algorithms publishes every two weeks. Each issue advances one argument, introduces one concept, or examines one case study from the emerging science of algorithmvigilance. It is written for clinicians — physicians, nurses, pharmacists, therapists, and allied health professionals — who use AI-assisted tools in their practice and want a rigorous framework for thinking about what those tools do after deployment.
It is not a tech newsletter. It is not a policy brief. It is a discipline in formation, published in public, one issue at a time.