PainSense is a structured-intake tool for complex health journeys. Patients with chronic or hard-to-categorize pain answer guided assessments over time; clinicians see longitudinal tracking that surfaces patterns the patient might miss in a 15-minute appointment.
In active build. The product surface is intentionally narrow — intake quality, longitudinal tracking, clinician handoff — because that's where the existing tools fail.
What it is
A mobile web app for patients, paired with a clinician view. Three things it does well:
- Guided intake. Branching questionnaires that adapt to the patient's answers, calibrated against actual clinical workflows rather than generic forms.
- Longitudinal tracking. Symptoms, medication response, daily impact — captured in a shape that's actually graphable over weeks and months.
- Clinician handoff. A summary view that takes 30 seconds to scan, surfacing the trends the clinician needs without forcing them to read every entry.
Why I built it
Anyone with a chronic condition has lived the version of healthcare where every appointment starts at zero. You re-explain your history. You re-describe your pain. The clinician makes notes that don't survive the handoff to their colleague. The "patient portal" is a 90s database in a 2010s skin.
PainSense is the version where the patient's own data is the canonical record — structured enough that a clinician can use it, owned by the patient, encrypted end-to-end so privacy isn't a checkbox.
How it ships
Built with Claude in the loop on every layer — intake form logic, encryption design, the clinician view. The HIPAA-aware design isn't a separate compliance pass on top; it's the architecture from day one.
Stack: mobile-first PWA (no native app store), client-side encryption for sensitive fields, a minimal server surface that never sees plaintext clinical content.
Why this matters for AI systems
Healthcare and AI share the same operator problem: the people doing the actual work — clinicians, on-call engineers — don't have time to translate a schema into their head. The interface either translates schema-to-operator-language, or the tool stops getting used.
PainSense is what happens when you take that discipline seriously in a regulated domain. Same posture I bring to AI ops surfaces: the operator's mental model is the source of truth; the data structure translates to it, never the other way.
Notes for verification (TODO before publishing): confirm the live surface description (mobile PWA, web app, native?), the actual tech choices, the encryption posture (HIPAA-aware vs. fully HIPAA-compliant — different bars). Replace before publishing.