Health Thread | Devpost

Inspiration

Healthcare often expects patients to remember every symptom, recognize important changes, and explain everything clearly during a short appointment. That can be especially difficult for people experiencing pain, fatigue, memory difficulties, language barriers, or difficulty speaking.

We built HealthThread to give patients more ways to communicate and turn scattered health information into a clear, patient-controlled story.

What it does

HealthThread helps patients capture, connect, and communicate their health experiences over time.

Patients can:

Describe a concern by speaking or typing
Point to the exact location of pain using an interactive body picture
Complete a guided check-in one question at a time
Review and confirm information before it is saved
Connect Fitbit measurements such as sleep, activity, and resting heart rate
Import previous ChatGPT, Claude, or Gemini conversations for review
Upload and organize existing health documents
View changes compared with their own personal baseline
Create an editable, patient-approved doctor summary
Have that approved summary spoken aloud in one consistent, natural voice
Let a doctor ask questions that are answered only from the patient’s recorded information
Use the experience in English or Spanish

HealthThread does not diagnose medical conditions. It helps patients preserve what happened, recognize meaningful changes, and communicate more confidently.

How we built it

HealthThread is built as a responsive web application using Next.js, React, TypeScript, and Tailwind CSS. Health information is managed through a local SQLite data layer.

We built separate Personal and fictional Demo profiles so real Fitbit information remains isolated from seeded demonstration data. The Fitbit connection uses OAuth and brings activity, sleep, and resting-heart-rate measurements into the patient timeline.

HealthThread combines patient-confirmed entries with wearable measurements to identify changes relative to the patient’s own history. Its summaries and doctor-facing answers remain grounded in confirmed records instead of inventing missing information.

Speech-to-text allows patients to record information naturally. Text-to-speech uses ElevenLabs to provide one consistent voice throughout the application. We also created an Expo Go mobile shell so HealthThread can be demonstrated as a phone experience.

Challenges we ran into

One major challenge was connecting information from very different sources without confusing measurements, patient statements, and AI-generated responses.

We designed a confirmation process so information does not silently enter a patient’s health record. Imported AI replies are never treated as patient facts, and possible health details remain separate until the patient approves them.

We also had to distinguish a real Personal profile from the fictional Alex demo profile, securely handle Fitbit authorization, build accessible voice interactions, and make the full experience work clearly on both desktop and mobile screens.

Another challenge was producing useful summaries without making medical claims. HealthThread is intentionally designed to organize and communicate information rather than diagnose.

Accomplishments that we're proud of

We are proud that HealthThread works as a connected experience rather than a collection of disconnected features.

A patient can speak, type, point to a body location, complete a guided check-in, combine those entries with Fitbit measurements, observe changes over time, and turn the resulting timeline into an approved doctor-ready explanation.

We are also proud of the application’s patient-control safeguards, bilingual experience, accessibility options, grounded doctor questions, and natural read-aloud voice.

What we learned

We learned that improving healthcare communication is not simply a matter of generating a better summary. Patients need flexible ways to express themselves, clear control over what is saved, and confidence that the system will not add or infer information without their approval.

We also learned how important personal baselines are. A change can be meaningful for one patient even when it would not appear unusual against a generic population average.

What's next for HealthThread

Our next step is to create consent-based live integrations with ChatGPT, Claude, Gemini, and other AI tools. A patient could approve a health detail while chatting and send only that approved information to HealthThread.

We also envision secure HIE and FHIR connections for clinical records, along with Apple Health and Health Connect integrations. These sources could join patient-entered information and wearable measurements in one patient-controlled timeline.

Our goal is for HealthThread to become the trusted connection between the tools patients already use and the healthcare professionals who care for them.

HealthThread — your health story, connected and ready to be heard.