Praxis, a standout project that won 1st Place Overall and Best Use of QNX at Hack the 6ix (with over 450 participating students), is a rehabilitation assessment platform built on QNX 8.0 that tackles a genuine question in healthcare: how do you measure whether a patient is actually getting better?

It's the kind of project that shows what happens when students get access to professional-grade, real-time tools like QNX Everywhere... They stop building demos and start building systems. For the team, the spark was surprisingly personal.

"One of the things that inspired Praxis was watching my grandparents' handwriting on their birthday cards slowly become larger and shakier year after year. It made me wonder whether we could use technology to actually measure those small changes in fine-motor control over time — and turn something you can only feel into something you can track." — Katherine Langille, Praxis team

That quiet observation became a bigger question: what if technology could track those subtle changes in fine-motor control over time?  It's a question that matters far beyond one family. Conditions like stroke, Parkinson's disease, and multiple sclerosis can all impair fine motor control. Patients are often prescribed tracing exercises to rebuild those skills, but between clinic visits there's rarely a quantitative way to tell whether performance is improving, holding steady, or slipping.

As the team saw it, a static drawing only tells half the story; it shows what someone drew, not how they moved while drawing it. The team set out to gather deeper insights by monitoring execution time and the number of tremors during the tasks. Praxis captures the full picture.

The team, Katherine Langille, Ananya Maheshwari, Sanvi Kaushik, AS Wasti, consists of two biomedical engineering students and two software engineering/computer science students, and deliberately aimed for the intersection of healthcare, embedded systems, and AI. The team composition clearly showcases the biomedical lens in what they measure and the software lens in how they measure it.

Rather than benchmarking against a general population, Praxis compares each session to the participant's own history across four metrics: mean deviation (how closely the trace follows the reference path), completion time, task completion percentage, and hand tremor.

The hardware is refreshingly accessible: QNX 8.0 running on a Raspberry Pi 5, paired with a camera and a small motion sensor (an IMU). The IMU rides alongside a pen, so it moves naturally with the participant's hand, while an overhead camera captures a clean, repeatable top-down image of each finished trace.

The workflow is sophisticated, yet simple. The sensor streams movement data throughout the task; when tracing finishes, the camera grabs a single high-resolution frame that's compared against the reference template. Critically, all of this runs on-device on the QNX real-time OS, so the assessment keeps working even with no network connection.

The team even ran a small language model locally on the QNX device to generate plain-language session summaries. Their smart guardrail: every number the AI reports is validated against the actual measurement, so it can add insight but can never alter a result.

From there, sessions flow to a backend and a React dashboard that visualizes trends over time. For deeper analysis, the team fine-tuned their own model, FreeSOLO, to explain whether a participant has improved, stayed stable, declined, or shown mixed results, with a Gemini-powered assistant helping less technical users navigate the dashboard.

Interestingly, Praxis didn't start on QNX at all. The team began with different hardware and changed direction largely because of how present and supportive the QNX team was throughout the event. They returned several times to talk through challenges and brainstorm solutions, and each conversation surfaced fresh ideas for bringing their vision to life in such a short window.

"[QNX’s] input really helped shape the final product, so it really was a collaboration," the team notes. That back-and-forth is exactly the kind of hands-on mentorship that turns a weekend prototype into something with real staying power.

Their biggest lessons came from iteration. When continuous pen-tracking hit occlusion problems, they redesigned the system for a single end-of-task image; simplifying the pipeline and boosting reliability. As they put it, a healthcare app takes far more than an accurate model: hardware, calibration, and data quality matter just as much.

That's the real magic of hands-on learning; you don't just read about building for safety-critical industries, you do it.

Congratulations to the Praxis team on a well-deserved win at Hack the 6ix.

Want to build your own project on a professional-grade RTOS? Explore what's possible with QNX Everywhere and start building today.