From patient behavior to clinical trial.

How do people with chronic conditions — diabetes, allergies — actually manage their disease day to day? Not in theory. Not in a clinical setting. In their homes, with their schedules, when they're tired, distracted, or just worn out by it all.

The question J&J was asking was whether intervention could genuinely change that behavior, or whether it would just add one more thing to manage. And a parallel question: why were patients making errors when self-administering biologics, and what would actually reduce those errors?

I led the research and design of AI-assisted coaching tools built around behavioral science and clinical inputs, working alongside J&J's clinical and R&D teams to turn what we learned about patients into service blueprints and experience strategies.

The tools weren't designed to replace clinical guidance — they were designed to meet patients in the moments between appointments, where most of the real disease management happens. That meant understanding what patients were actually doing, not what the care plan assumed they were doing.

In parallel, I redesigned the instructions for self-administered biologics for children. The printed materials weren't working — errors were frequent and the consequences were significant. The question was what format would actually reduce those errors under the real conditions people were using them: at home, often alone, sometimes under stress.

Most health interventions are designed around what clinicians want patients to do. That's not the same thing as what patients are capable of doing — consistently, in real conditions, over time.

Adherence isn't a willpower problem. It's a design problem. The environment, the format, the moment, the cognitive load — all of it shapes behavior more than intention does. When you design for that reality instead of the ideal version, outcomes change.

The numbers here bear that out clearly.