Apple Fitness+ service has been a hit with the users. Apple’s unique ability to integrate Apple Watch metrics automatically has been a huge success for Fitness+ users. Apple is upping its offerings for Fitness+.
According to a new Apple Patent that was approved in January 2021, “TECHNIQUES FOR PROVIDING CUSTOMIZED EXERCISE-RELATED RECOMMENDATIONS,” Apple is already looking at taking its Fitness+ service to the next level by incorporating mechanisms to provide unique exercise recommendations to users based on several factors.
Although health and fitness devices provide some feedback today, these devices do not provide the level of customized support and advice that one would receive from a personal trainer.
Similarly, exercise machines such as stationary bikes, weight lifting machines, treadmill machines, elliptical machines, and the like often provide exercise feedback to the user during an exercise session.
However, such feedback is often transitory or underutilized in conventional health and fitness systems.
That is one of the reasons why Apple is developing its own proprietary exercise recommendation engine.
The patent highlights examples of some recommendations that could be provided to the user.
These look like both exercise and health-related recommendations (e.g., “drink more water,” “ingest more protein,” “get more sleep,” “reduce your sugar intake,” etc.) or any other type of recommendations.
An “exercise recommendation” can include suggestions to use a different type of exercise machine.
For example, it may suggest the user do more cardio via an exercise bike instead of doing core workouts.
An exercise recommendation could also suggest a change in speed, weight, repetition, form, type of exercise machine, or any suitable characteristic related to a user’s utilization of one or more exercise machines.
In at least some examples, an exercise machine may include (or be otherwise in communication with) one or more sensors (e.g., accelerometers, gyroscopes, cameras, blood pressure sensors, oxygen sensors, heartrate sensors, thermometers, etc.).
Inputs to the Exercise Recommendation Engine
The Exercise recommendation engine will consider a user’s workout history, user’s past vitals, and current health metrics and use sophisticated algorithms to generate specific and actionable recommendations.
The model described in the paper also talks about taking user profile information into account that could contain a user’s medical information such as current medication being taken, the disposition to a particular disorder, or more.
There is also the component for goals. Some users would like to train for a 5k or achieve a certain weight loss goal. Fitness+ recco engine would consider the goals as well.
The algorithms will use sophisticated classification models that will utilize suitable unsupervised machine-learning techniques, such as cluster analysis algorithms (e.g., k-means, etc.), to identify similar users.
Execution of a cluster analysis algorithm may cause a set of objects (e.g., users, fitness-related information) to be grouped in a way that objects of the same group (called a “cluster”) are more similar to each other than users of other groups (clusters).
This could lead to better benchmark information. Think of a more robust and enriched burn bar in the future.
The patent, 0210001180, was filed recently on September 18th, 2020, and approved today.
Inventors on the patent include Robert Pitchford, SW engineering Manager at Apple, Stephen Holter, Sr. Manager, Wireless design, Apple, and Wang Ying.
Apple is not planning to upload exercise videos and offer new types of exercises with various trainers.
The Cupertino thinking is to take it up a notch using machine learning models to figure out what exercises work best for you and your goals and offer you customized health and exercise recommendations based on your history and aspirations.