From fitness apps trying to do more than count steps to nutrition trackers - how Indian health start-ups are using AI for fitness

AI for wellness: From an app that can tell you the nutritional value of a plate of food from a photo, to a personalised AI-driven fitness coach - AI-led wellness firms are trying many different combinations as LLMs become more focused and AI becomes accessible.

What's artificial intelligence (AI) got to do with wellness? Consider this: nutrition and fitness apps such as HealthifyMe, Fitelo and Fitterfly have already been using AI for years now.

HealthifyMe has used AI to analyse data from its users’ continuous glucose monitors (CGM) to develop a virtual CGM to predict glucose spikes and drops. Fitelo leverages AI to create data-driven and personalised solutions. And Fitterfly uses AI to help clients manage lifestyle diseases such as diabetes.

“By delving into individual user data, AI creates user-engaging wellness programmes,” said Sahil Bansal, co-founder and CEO, Fitelo.

Companies are also developing AI-powered chatbots to answer queries, and fitness practioners are using tools like ChatGPT to create personalised diet and exercise plans, noted Dr Arbinder Singal, CEO and co-founder, Fitterfly.

However, these applications are relatively straightforward; AI has the potential to accomplish more. AI is about to transform healthcare with personalised coaching and multi-modal engagement, said Tushar Vashisht, co-founder and CEO, HealthifyMe.

The early signs are already there: activity tracker Whoop! launched its own AI coach in partnership with Open AI and the Google-owned Fitbit's AI-powered device is set to launch soon.

“Most of these products are unidimensional and focus only on heart rate, sleep or steps. There is a real possibility of somebody building a comprehensive AI coach that takes into account all your multiple data vectors and makes sense of that, analyses them and responds,” said Vashisht.

AI for wellness

As the amount of data available grows and expansive language learning models learn more, AI will go beyond analysing your steps, sleep and heart rate and evolve into a coach that also takes into account your capabilities, nutrition, dietary restrictions, health parameters, disease history, stress and emotional state.

“Imagine a coach that knows a lot about your nutrition, fitness, stress, sleep, and responds to your data vector in a very personalised way… who knows you and your background quite well. And not just the structure data, but also whether you have an injury, your emotional consideration, and more. We are certainly working on a coach like this who can act as your lifestyle co-pilot,” added Vashisht.

“In personalised coaching, we want human coaches to be able to feedback and hold us accountable. AI can play a huge role in that by enabling and empowering these human coaches. We are working on a co-pilot for our human coaches that can help them do their jobs very efficiently,” Vashisht added.

AI for monitoring health, faster diagnosis

The volume of data generated by reports, CGM devices, fitness trackers, and the like exceeds the processing capacity of a human brain, said Fitterfly’s Singal. The biggest advantage that AI holds over humans is the tremendous speed with which it analyses these vast amounts of data that are coming in and also collected.

“The real-time monitoring capabilities of AI help in redefining the way health metrics are tracked. AI continually monitors health data and dynamically adjusts plans for optimum results, up to 500 percent faster than traditional manual methods. This ensures a proactive approach to individual health progress,” said Fitelo’s Bansal.

The world has already started moving beyond text-based AI to include voice, pictures and videos. “We anticipate a shift to voice-based interactions within the next two years. Additionally, AI is poised to generate humanoid videos, establishing emotional connections with individuals grappling with chronic diseases or aspiring to enhance their fitness,” said Singal.

Vashisht pointed to the use of photographs by AI for things like calorie counting and nutrition tracking. “We launched Snap a little while ago. It allows for automatic nutrition tracking... You can track a photo (of your food) as long as you give us gallery permissions. We convert that picture into nutrition values for consumers,” added Vashisht.

Role of human healthcare providers

As AI evolves, it is expected to become so sophisticated that end users may not discern its presence in the background at all, contended Singal.

Recent trends indicate a positive shift in user attitudes. “People are willingly engaging with AI, which is evident in the utilisation of tools like ChatGPT for weight loss, with an understanding that it operates on AI principles. This trend is likely to grow, fostering increased engagement with AI models across various domains,” said Singal.

Personal data is used to analyse health and improve wellness plans, and some users might be apprehensive about sharing personal data, admitted Bansal. India is among the top five countries in terms of investments received by AI startups. But the country faces certain challenges in harnessing the full potential of the technology. “AI systems can be biased if they are trained on data that is not representative of the population they are used to serve,” said Bansal. To make this work long-term, he added, we need people to share data, and for that we need data protection protocols, encryption, anonymisation of data, and adherence to privacy laws by organisations.

Singal added that AI can't replace higher-skilled professionals such as doctors and nurses, but “the integration of AI can lead to improved diagnoses and outcomes. Tasks like coaching... are also becoming automated. In this scenario, the role of a human coach will be to empathise, nurture, and guide patients toward better outcomes, as these are elements that AI may not fully replicate”.