OnTracka
High-compliance infrastructure for clinical trials and real-world patient monitoring — with an AI/RAG layer for researchers.
Overview
OnTracka is a purpose-built platform for investigator-led and sponsor-led clinical trials, supporting structured endpoint capture, longitudinal monitoring, and audit-ready workflows in regulated care environments. As backend architect and full-stack lead, I built the platform's core from the ground up.
Trials still lean on paper diaries, spreadsheets, and email. Patient-reported data arrives late, incomplete, or recalled from memory; sites only discover who has fallen behind at the next visit; and side effects surface too slowly to act on. What does get collected is rarely structured enough to stand up to an audit.
OnTracka puts the whole study on one connected system. Participants report from their phone on the protocol's schedule, with wearable and health-app data syncing automatically. Clinicians watch adherence and safety signals across the cohort in real time and run telehealth visits without leaving the record. Every reading lands as structured, timestamped, audit-ready evidence — and a RAG layer lets researchers query the dataset in plain language.
What I did
- Engineered the core backend architecture for secure collection and scheduling of complex patient health data across clinical research trials.
- Built the web portal for doctors, research analysts, and administrators — real-time data visualization and patient-progress monitoring.
- Implemented time-sensitive data pipelines that automate patient self-monitoring and real-world studies against strict clinical schedules.
- Designed and shipped a retrieval-augmented generation (RAG) system on the OpenAI Agents SDK so researchers can securely query and analyze patient datasets through an AI interface.
How the three products fit together
OnTracka is three connected surfaces: a portal where clinicians and researchers analyze participant data, a web app that onboards patients onto a study, and a mobile app participants use to run the study day to day. Here is the flow, screen by screen.
Cohort command center
The dashboard clinicians and researchers land on — live adherence across the whole cohort, with patients on-track, pending, and missed, plus per-baseline-day completion and adverse-event signals.
Intake form
Drilling into a single participant. The multi-part intake captures cancer type, stage, and current treatments — the clinical baseline every later reading is measured against.
Medication tracking
Every product and dose the participant logs, ranked by frequency over the selected window — so clinicians see exactly what's being taken and how often.
Protocol & patient-reported outcomes
The study-protocol timeline (Baseline → Day 84) with validated PROMIS and PRO-CTCAE questionnaire responses and running adherence — the structured, audit-ready record a trial depends on.
Clinician scoring & notes
Clinicians score anxiety, depression, stress, pain, fatigue, and focus with trend deltas, and keep threaded clinical notes against the participant.
Longitudinal symptom trends
A visual summary that plots any tracked symptom over time, turning weeks of self-reports into a single readable curve.
Video visit — clinician side
Telehealth is built in. A clinician runs a live video consult while taking structured notes — side-effects questionnaire, observer rating scale, checklist — without leaving the participant record.
The same visit — participant side
The other end of that call, on the participant's phone — one-tap video with their clinician, and a quick rating prompt once the session ends.
Adverse-event self-report
A guided flow for participants to flag unusual side effects — describe them, rate severity, note whether they're new or worsening, and how disruptive they are — with a clear notice that it isn't real-time monitoring.
Connect health data
Part of onboarding onto a study: participants securely link Apple Health, Google Health, or Fitbit so medication effects can be correlated with sleep, heart rate, and activity.
Daily home
The participant's day-to-day hub — daily reporting prompts, symptom trends, medication progress, and synced wearable data (HRV, steps, resting HR, sleep) in one place.
Health insights
Personal insights that keep participants engaged between study touchpoints — zone minutes, heart-rate and sleep trends, and encouraging weekly-goal feedback.
Building something similar?
I can own it end-to-end or drop into your team.