Oleg Lola, founder and CEO at MobiDev, a custom software development company.
AI has been a part of the fitness software market for over 10 years, but some fitness industry leaders are still confused as to how to incorporate AI into their fitness applications. With the capabilities of AI changing and expanding quickly, it can be difficult to keep up with what’s possible—and not possible—for AI in fitness. Today, I want to provide a synopsis of what current AI technology can provide to makers and users of fitness apps while touching on current challenges, as well.
How AI Accelerates The Fitness Industry
Modern AI-enabled fitness approaches are typically divided into two categories: algorithms that support athletes with workout programming, fatigue tracking, calorie count, sleep pattern recognition, etc.; and motion analysis algorithms that improve performance by suggesting technique changes or coaching various movements. Both can be used to enhance fitness apps for users.
Human Pose Estimation
For a long time, movement tracking was only available in smart wearables because computer vision wasn’t advanced enough to provide real-time tracking. Today, AI-based human pose estimation technology has allowed the analysis of the position of a body based on visual input without wearables. This can be used in a number of applications such as movement analytics apps that help athletes improve performance in specific movements. Human pose estimation would, for example, perform real-time analysis of users’ exercise recordings and provide suggestions on how to improve exercise form and biomechanics..
Personalized Training Programming
Personalized training programming can use AI-based machine learning to create personalized diet programming, sleep and rest periodization, workout load periodization and more utilizing user data input. In a world where consumers increasingly demand greater levels of personalization, machine learning-based personalized recommendations are likely to become the standard for fitness apps in the future.
All-In-One Coaching Bots
Coaching bots that combine multiple types of AI such as human pose estimation and conversational AI can support athletes in perfecting their exercise technique, posture correction, exercise choice, diet recommendations and more. A comprehensive coaching bot that utilizes AI brings the cutting edge of fitness technology together with the convenience of an all-in-one app experience.
Injury Prevention And Rehabilitation Apps
Injury prevention and rehabilitation apps may use human pose estimation to scan movements for possible risk of injury and for rehab purposes. 3D pose estimation models can provide accurate enough tracking to support athletes in precise rehabilitation exercises for muscle imbalance correction as well as ligament and joint injury treatment.
Challenges Of Fitness AI
While fitness-enabled AI apps have made great strides toward better user experience and data analysis, they still have some challenges that app developers need to address to keep the industry moving forward.
Technical Limitations
Any AI app is highly dependent on the data provided. As it can be difficult to input information into the workout routine of a single person or a sports team from an end-user standpoint, this can be a headache for users.
For instance, one major difficulty of motion analytics is that the user must follow strict rules while filming themselves to achieve acceptable results. Without following specific rules, they may be unable to capture proper video that provides enough input to the model. Things like poor lighting conditions, inappropriate camera angles, the presence of multiple people in the frame and occlusive objects such as clothes or gym racks can result in unusable video and multiple frustrating retakes.
Data Accessibility
When it comes to fitness, there are thousands of information points that developers want the models to recognize, including scores of different body dimensions, sleeping patterns and training capabilities. All of this requires an immense amount of data collection. Until this data is gathered, AI-enabled fitness apps can only achieve so much.
Is AI Trustworthy?
When we rely on AI for decision making, one pressing problem is that we don’t actually know how AI makes its conclusions and recommendations, and yet we fully delegate responsibility to it. The ethics and reasoning behind AI coaching may raise the question, “Can we trust what AI is telling us?” Product-team wise, it will always be a question of how to make applications more trustworthy to users, since artificial intelligence can potentially bring harm if there is not enough explanation behind its decisions.
Privacy And Regulations
Not all data can be collected and stored without regulatory burden. When AI-enabled fitness apps communicate and store information about diet programming or rehab procedures, that opens up a whole new world of healthcare regulations. The wider the scope of the AI fitness project, the more privacy concerns it will encompass.
The Future Of AI-Enabled Fitness Apps Is Bright
With AI constantly improving and becoming more popular, the world of AI-enabled fitness apps will likely only grow. If developers can solve the inherent challenges of AI, while maintaining and improving the functions of these apps that users already know and love, there is no limit to what AI-enabled fitness apps can do in the future.
Forbes Technology Council is an invitation-only community for world-class CIOs, CTOs and technology executives. Do I qualify?
