Fit Pat
An Android fitness app that starts from your health condition rather than your workout. Pick high blood pressure, back pain, diabetes or obesity, and every exercise after that is filtered by the answer.
The problem
Fitness apps assume you are healthy. If you are managing high blood pressure, diabetes, back pain or obesity, the exercise you need is different from the exercise the app gives everyone, and none of the popular ones ask. The alternative is booking a doctor to be told which movements are safe, and then waiting.
That is the gap Fit Pat aims at. Not another workout tracker, but an answer to a narrower question: what can I safely do, given this condition?
My approach
The condition comes first, not the workout. The opening screen asks what you are managing, and everything downstream is filtered by that answer: condition, then activity type, then a planned set of exercises with repetitions, duration and a demonstration video.
I mapped conditions to exercises explicitly. Each condition is its own activity class (HighBloodPressure, BackPain, Diabetes), and each pairing of condition and activity is another, such as YogaDiabetes or StretchBackPain. That does not scale past a handful of conditions, and today I would hold the mapping as data rather than as classes. But it was explicit and easy to check, which mattered more than elegance for content where a wrong suggestion could hurt someone.
Pictures instead of words. I used illustrations and animation rather than text and buttons throughout, because the users I had in mind, older adults managing a condition, are not necessarily fluent with apps. That was the most deliberate design decision in the project.
Around that core flow sit a pedometer built on the device's motion sensor, a route tracker for walks and rides, and a BMI calculator, with accounts stored locally in SQLite.
The outcome
The app runs on device: sign-up and login with password reset, four conditions, activity dashboards, exercise plans with embedded video, step counting, route mapping and BMI.
I listed the weaknesses in my own proposal and they turned out to be the real ones: tracking errors, and a pedometer whose readings are not always accurate. A step counter built on raw motion-sensor events over-counts on rough ground and under-counts a slow walk, and I never solved that.
The honest limitation is larger than accuracy. The mapping from condition to exercise was mine, assembled from research rather than from a clinician. For an app whose entire premise is that the movements are safe for your condition, that is the part which would need professional sign-off before anyone should rely on it.
