Apple Invents a Personal Trainer Level Exercise Guidance App that could use Adaptable Machine Learning
Today Patently Apple discovered a new patent application from Apple in Europe that relates to extensive exercise programs that could match the quality of a personal trainer once set up. In one example, Apple notes that a machine learning methodology could be employed in developing a detailed plan that will track a user's progress using an array of exercise equipment at a gym.
Apple's Patent Background
As individuals are devoting more and more time to non-physical activities, and as the rates of obesity and disease are on the rise, the need for greater amounts of physical exercise is a reality for many. It is often difficult for a person to educate themselves with respect to proper exercise, as conflicting information is widespread.
For those with the financial means and desire, a personal trainer or a health coach can be excellent resource for achieving fitness goals. However, the costs associated with utilizing a personal trainer or health coach may be prohibitive for many people.
Many individuals are turning to health and fitness devices to help track their health and fitness progress. Although health and fitness devices provide some feedback, these devices do not provide the level of customized support and advice that one would receive from a personal trainer, as such devices are often limited to processing particular types of data (e.g., feedback from onboard sensors, user input, etc.).
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.
Apple's invention covers systems, methods, and computer-readable medium, for providing customized exercise-related recommendations.
A computer-implemented method may comprise identifying, by a service provider processing device, user interactions with a particular exercise machine or a plurality of exercise machines, the user interactions associated with a user profile, each of the plurality of exercise machines comprising a computing device in communication with the server provider processing device.
The method may further comprise obtaining, by the service provider processing device, exercise information corresponding to a plurality of user profiles, the exercise information being obtained from the plurality of exercise machines.
The method may further comprise generating, by the service provider processing device, a classification model based on the exercise information corresponding to the plurality of user profiles. The method may further comprise receiving, by the service provider processing device, exercise metrics associated with the user interactions with the particular exercise machine.
The System: The system may comprise a plurality of exercise machines individually configured with one or more sensors, one or more data networks, one or more processors, and one or more memories storing computer-readable instructions.
Executing the instructions (by the one or more processors) may cause the system to at least collect, at an exercise machine, exercise information for a plurality of user workout sessions, wherein the exercise information comprises a duration of a workout, a level of workout, or a number of repetitions performed.
Executing the instructions may further cause the system to identify a first user workout session of the plurality of user workout sessions. Executing the instructions may further cause the system to identify first user information corresponding to the first user workout session.
Executing the instructions may further cause the system to access the collected exercise information. Executing the instructions may further cause the system to identify a plurality of exercise recommendations based on the first user information and the collected exercise information. Executing the instructions may further cause the system to determine a customized exercise recommendation from, a plurality of recommendations based on the first user information and a comparison between the collected exercise information and expected progress data. Executing the instructions may further cause the system to present the customized exercise recommendation.
Apple's patent FIG. 1 below is a simplified block diagram illustrating an example flow for providing a customized exercise-related recommendation.
Apple's patent FIG. 3 below is a simplified block diagram illustrating an example flow for providing a number of customized exercise-related recommendations.
Apple's patent FIG. 2 below is a simplified block diagram illustrating an example flow for maintaining a machine-learning model utilized to provide customized exercise-related recommendations.
Machine Learning Training Program
On Machine Learning Apple specifically states the following:
"In at least one embodiment, a classification model may utilize any suitable unsupervised machine-learning techniques, such as cluster analysis algorithms to identify users that are similar to one another.
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).
In a non-limiting example, cluster analysis may be utilized to classify previously unclassified historical data, where user/fitness-related information is clustered into a suitable number of clusters and each cluster is assigned a classification. The assigned classifications may be then utilized as training data to train a classification model with a supervised machine-learning algorithm.
Subsequent fitness-related information may be added to the training data set, and the classification model may be retrained on the new training data set at any suitable time."
Apple's patent application was originally filed back in March 2018 and published yesterday in Europe. Considering that this is a patent application, the timing of such a product to market is unknown at this time.
Apple inventors include Steven Holter, Sr. Manager, Wireless Design; Robert Pitchford, Software Engineering Manager; and Ying Wang, Software and Wireless Design Engineer who developed an innovative robotic system for wireless design. Wang left Apple in late 2017 to work at Verizon.