Fit2apple

🏋️‍♂️ About the Project – Fit2Apple

🔥 Inspiration

While working on fitness app integrations, I noticed a common problem: many apps log the wrong workout type in Apple Health. Whether it’s logging a HIIT session as "Other" or a strength workout as "Functional Training", these misclassifications break the user experience — especially for those who rely on accurate fitness tracking. I built Fit2Apple to solve this.

💡 What I Learned

• How Apple HealthKit handles HKWorkoutActivityType and the importance of accurate classification for sensor tracking and user insights
• The diversity of workout formats and how users and apps interpret them differently
• Developers need simple, pluggable APIs they can trust without adding manual mapping logic

🛠️ How I Built It

• Tech Stack: Node.js, AWS Lambda, API Gateway, DynamoDB, and RapidAPI
• Designed a 4-stage data pipeline:

1. Normalization of workout input (e.g., reps, sets, rounds, intervals)
2. Heuristic rules & inference based on structure, pacing, and intensity
3. Activity type classification into the most accurate HKWorkoutActivityType
4. Return metadata to help developers understand the classification reasoning
   • Built a public demo: fit2user.vercel.app

🚧 Challenges

• Apple provides no public documentation or guidance for how to choose HKWorkoutActivityType based on custom workouts
• Workout formats are inconsistent across platforms — from circuits to Tabata to freestyle
• Developers rarely think about the Apple HealthKit implications until users complain
• Finding the balance between too simple and too complex was tricky — I had to support edge cases while keeping the API intuitive

Built With

• node.js