As a society, we are increasingly focused on health and wellbeing with influences from the media and professionals ever present. This has prompted a huge rise in wearable technologies as a method of keeping track of elements such as activity level and calorie intake. With their built in reminders and personalised features, products such as the Apple Watch, FitBit, and Garmin have opened a new market for technology.
Fitness apps compatible with the hardware are increasingly popular, but what makes each fitness products stand out from the crowd? And how are you supposed to stick to your chosen platform once you’ve signed up?
We spoke to Freeletics, the high intensity interval training app that combines full body routines and exercises with running for ‘a complete fat shredding workout’, ahead of their talk at the Machine Intelligence Summit Amsterdam 28 & 29 June. The app, which is compatible with most wearable technologies, concentrates on motivating its users by creating ‘a community of athletes’ who can then compare their progress against their friends on the app.
To personalise these workouts and make them tailored to each individual, a significant amount of programming has to take place. Freeletics are using AI and more specifically machine learning to train the product which is now ‘actively learning from all 14 million users’ performance and progressing with them over time’. The ‘coach’ in the app takes into consideration ‘dozens of variables and compares them with those of other users.' This intelligent feature also helps to reduce the risk of injury or over training by analysing your workouts against your fitness level amongst other features.
Laith Alkurdi, Senior Machine Learning Engineer at Freeletics told us how his mission is to ‘optimise the users’ training experience through the coaching algorithm’. In his role, Alkurdi is required to go through the entire data pipeline every day to address different data-driven hypothesis as well as design prediction algorithms based on sport-scientific and psychological concepts.
We were interested to hear more about Laith's role so asked him a few questions about his current work:
Can you give us a short overview of your work at Freeletics?
Every project starts with data preparation, which usually consists of identifying with the relevant data sources, alongside data warehouse developers, and follows with data exploration and building first unsupervised learning algorithms; while taking input from the relevant domain experts. Data is then represented in the best manner to produce features that feed into the optimisation step. In this optimisation step, parameters are learned and the resulting algorithm is evaluated based on user training behaviour. Additionally, part of my responsibilities is communication across the different functions of the company and explain how the algorithm works and the ranges of benefits that could be communicated to the users.
What do you feel are the leading factors enabling recent advancements and uptake for machine learning for personal training?
Maintaining a healthy lifestyle and adhering to a training program is a notoriously challenging task. From statistics, only one in 10 succeed in keeping their health related new year resolutions. Two major issues can be hypothesised here as to why people fail in adhering to general fitness programs. The first being lack of guidance, while the second could be attributed to the fact that fitness programs generally serve the needs of the average person and fail to take their personal needs into account. Machine learning comes to address exactly these two issues. Here at freeletics we aim to be the personal coach in our user's pocket guiding them and studying their performance towards prescribing the best workouts that fit each user's goals.
What impacts on progress did you observe by using machine learning to create personalised fitness programs?
The biggest themes here can be clustered around personalisation, adaptation and adherence. A successful Machine learning implementation should optimise these three major metrics regarding training experience and fitness programs.
What developments can we expect to see in machine intelligence in sports in the next 5 years?
As biosensors become cheaper, more accurate and more accessible Machine intelligence would benefit for this added stream of data to make even more in-depth analysis and more personalised suggestions based on the user's needs. Heart rate sensors coupled with computer vision algorithms could mean that you have all the benefit of a personal trainer right at your side at all times!
|Register for the Machine Intelligence Summit in Amsterdam, on 28 & 29 June, where Laith Alkurdi will be giving a high-level description on how the Coaching A.I. makes use of the multiple interaction points within the Freeletics ecosystems and how it leverages itself to be each athlete's personal coach.|
21 September 2017, London
The Deep Learning Summit is at the forefront of AI. Explore the impact of image & speech recognition as a disruptive trend in business and industry. How can multiple levels of representation and abstraction help to make sense of data such as images, sound, and text. Hear the latest insights and technology advancements from industry leaders, startups and researchers.
21 September 2017, London
The next generation in predictive intelligence. Anticipating user & business needs to alert & advise logical steps to increase efficiency. The summit will showcase the opportunities of advancing trends in AI Assistants & their impact on business & society. What impact will predictive intelligence have on business efficiency & personal organization?
21 September 2017, London
Following day 1 of the summit, attendees will come together for an evening of networking, discussions and fine food & wine. Mix with leaders on topics including NLP, speech recognition, reinforcement learning and image analysis, as well as applications in sectors including manufacturing, transport, healthcare, finance and security.