Healing Hands

Inspiration

Healing Hands started with my own recovery experience after a car accident. I fractured my hands, and when the casts came off, I could barely move my fingers. I knew I needed to keep exercising, but “practice more” wasn’t enough guidance once I was back home.

What helped most was having a therapist. However, therapy was expensive and difficult to access from home, and traveling to frequent appointments was especially challenging while I was injured and had limited mobility. If anything unusual happened between appointments—or if I made small improvements—it was difficult to notice on my own.

That experience gave me the idea for Healing Hands: a camera-based tool that gives people clear feedback while they practice between appointments. It does not replace a therapist or make medical decisions. It simply helps users follow an exercise, complete it consistently, and keep a record of what they did.

Why it matters

After researching the topic, I found that hand and wrist injuries are much more serious than most people realize:

10–30% of emergency-room visits involve the hand or wrist.
Around 50% of patients stop doing their home exercises.
In the cost data we reviewed, hand and wrist injuries ranked #1, at approximately $740 million, compared with $532 million for hip fractures.

Treatment does not end when the cast comes off. Recovery often depends on repetitive exercises performed at home—but that is where people have the least feedback and motivation.

When exercises feel painful, repetitive, or disconnected from visible progress, people may stop practicing. That can contribute to continued stiffness, weakness, reduced movement, and a more difficult recovery.

I'm not alone. Healing Hands focuses on this overlooked gap: helping people stay engaged between appointments by making every movement visible, responsive, and rewarding.

What it does

Healing Hands is a camera-based hand-training companion that adapts to how each person can move today.

The user begins with a short calibration and performs three movements:

Making a comfortable fist
Touching the thumb to each fingertip
Bringing both palms together and holding

Healing Hands uses the person’s current movement as a baseline instead of comparing everyone with the same fixed standard.

During training, the camera tracks the hand and turns each exercise into an interactive experience. Particles react to movement, fingertip touches create musical feedback, hold exercises build visible progress, and completed repetitions trigger rewarding effects.

This creates a simple recovery loop:

Move → receive feedback → adjust → try again → recognize progress.

How we built it

We started with the recovery experience rather than the technology: injury, difficult home practice, missing feedback, and invisible progress.

We designed three exercises to demonstrate how calibration, personalization, and playful feedback could make practice more engaging. The browser observes 21 points on each hand and follows the movement over time.

To avoid counting accidental movements, Healing Hands checks sequence, stability, hold duration, tracking confidence, and whether the user releases before beginning another repetition.

We then connected those movements to particles, sound, progress indicators, and scoring. Codex and GPT-5.6 helped us develop the concept, build the prototype, solve tracking and camera problems, refine the experience, and create the animated product story.

Challenges we faced

The biggest challenge was making imperfect camera tracking feel reliable. Fingers overlap, lighting changes, and hands disappear—especially during two-hand exercises—so we had to track movement over time, not just frame by frame.

We also had to make recovery feel enjoyable without making it feel less serious. Most importantly, Healing Hands adapts to each person’s current ability, so limited movement still feels like progress—not failure.

What we learned

We learned that tracking a hand isn’t enough—the system must understand the full movement and know what guidance to give next. Immediate feedback makes small improvements visible and repetitive practice more motivating. Most importantly, personalization should shape each user’s goals, measure their progress, and adapt over time rather than applying the same fixed standard to everyone

What’s next

Our ambition is to grow Healing Hands into an accessible recovery platform that connects people, professionals, and intelligent personalized guidance.

Professionals could review progress between appointments and adjust plans, while the system adapts daily difficulty according to the person’s movement and consistency.

The same approach could expand beyond hands to wrist, arm, shoulder, balance, and mobility training. Exercises could become interactive worlds involving music, games, virtual gardens, and gentle challenges selected around each person’s interests.

At a larger scale, Healing Hands could become a human-in-the-loop simulation environment—not only collecting real movement data, but also calibrating simulations, filling training gaps, and creating a continuous cycle of action and real-time human feedback. This loop could improve physical AI models and train rehabilitation robots to understand and adapt to real human movement.

Our larger vision is to make guided recovery available wherever someone has a camera—turning isolated home exercises into a continuous, personalized, and encouraging recovery journey.