Smart Registration of Clinical Interventions: Hackathon IACC 2026

1. The Clinical Challenge in 72 Hours

During the Hackathon IACC 2026, my team (consisting of Karen, Patricio, and myself, acting as Tech Lead and representative) collaborated to solve a real-world challenge for Fundación Cristo Vive, specifically for their Talita Kum addiction rehabilitation center.

The core problem was that professionals needed to record therapeutic interventions (such as relapse alerts or emotional emergencies) in a matter of seconds, in corridors, without interrupting their support process, and often without internet access. Lacking tools adapted to this environment, the information was documented late, incompletely, or lost altogether.

The goal was to design a mobile application (PWA) that allowed capturing and structuring this clinical information quickly through voice, guaranteeing data security.

Project Demo in Action

Below is the real-time demo of the application (voice dictation, structured AI output, and automatic secure local cache purging):

2. Architecture and Technology Stack

To guarantee speed, on-site availability, and security, I designed the following architecture:

Frontend & Backend: Next.js (App Router) and Tailwind CSS, structured as a Progressive Web App (PWA) to function as a native app on the therapists’ mobile phones.
Edge Database: Turso (libSQL) managed with Drizzle ORM. This allowed us to achieve extremely high response speeds.
Authentication and Roles: We implemented Better Auth to strictly segment access. The Therapist Role can only record and validate their interventions, while the Administrator Role accesses a Dashboard with KPIs and strategic indicators for the management.

3. Artificial Intelligence Integration and the “Golden Rule”

The core of the solution was voice capture processed by the OpenAI API.

Whisper Transcription: The therapist dictates what happened, and the system converts the audio to plain text.
GPT-4o-mini Structuring: Using Structured Outputs (JSON), the model extracts the information and classifies it exactly into the 5 mandatory fields required by the institution: objective of the intervention, details of what occurred, agreements, follow-up actions, and relevant observations.

Human-in-the-loop (The Golden Rule): The hackathon rules required that technology facilitate recording but never replace professional judgment. Therefore, the AI only pre-fills the form. The therapist is required to review, edit, and validate the info before saving it permanently.

4. Cybersecurity and Ethics (The Greatest Technical Challenge)

Since we were handling mental health data, security had to be integrated into the design from day one. During development, and after iterating on feedback from our advisor professor, I implemented three critical measures:

Mandatory Verbal Consent: I added a UX protocol (a mandatory checkbox) prior to recording, which forces the therapist to confirm that they obtained verbal consent if the patient/user is present, safeguarding the ethics of the process.
Risk Mitigation (Offline-First): To guarantee operational continuity, if there is no internet connection in the corridor, the data is temporarily saved on the device using IndexedDB and Service Workers.
Secure Cache Destruction: To mitigate the risk of loss or theft of the therapist’s mobile phone, I programmed the local cache to be automatically deleted once the device recovers its connection and syncs the data with the Turso database, ensuring no sensitive clinical data remains on the device.

5. Results: 🥈 Second Place

On July 8th, 2026, we received the official results: we placed second in the Hackathon IACC 2026.

This was an achievement the entire team (Karen, Patricio, and I) felt deeply proud of. The dedication across 72 hours, the architectural decisions made under pressure, and the ethical approach we brought to the project were recognized by the jury of IACC and Fundación Cristo Vive.

As recognition, the team will receive gift cards and Python certification vouchers for each member. We were also invited to attend the awards ceremony at the Talita Kum center — the very place where the challenge we solved was born.

Beyond the result itself, the experience of building a real-world solution with direct impact on the daily work of mental health professionals was the greatest learning of the entire process.