SDG 3: Good Health and Well-being
2. Project Details
Company or Institution
Using AI to Accelerate Changing the Eating Habits of a Nation
General description of the AI solution
We will work with Dr. Thanh Nguyen, a well-known AI expert from University of Oregon, to design and manage the AI aspects. We aim to use AI to improve the effectiveness of our MyThali social media program by: (i) finding the best influencers in social networks of urban consumers in India; and (ii) figuring out the optimum number of times per week that selected influencers should communicate with their peers on social networks.
Essentially, we expect these social influencers to help us in: (i) drawing attention to our MyThali dietary education program; and (ii) influencing network members to improve their eating habits. Based on the social network data, we will identify a subset of the most influential members. We propose to explore AI methods in Influence Maximization Research to determine these influencers, by measuring the ultimate impact of each influencer not only on their direct peers (eg followers) but also on other members in the entire network, through the computational analysis of how information can be spread on social networks. In addition, given the uncertainty about the network topology, we plan to apply techniques in Graph Learning such as Link Prediction to generate a more accurate view of the network and predict hidden connections among members using machine learning techniques such as deep neural nets.
Furthermore, given the dynamic nature of social networks and human behavior, we aim to determine the optimal action plan (ie when and how often each influencer should communicate with his/her peers on social networks) to maximize the impact of our program. Repeatedly spreading dietary information can create long-term memory for recipients but may be burdensome for influencers. To cope with this problem, we will explore techniques in Multi-Agent Systems and Reinforcement Learning that model human strategic interaction and decision-making process over time on social networks.
Excellence and Scientific Quality: Please detail the improvements made by the nominee or the nominees’ team or yourself if your applying for the award, and why they have been a success.
In this project, we will use techniques from multiple sub-fields of AI. More specifically, we will apply methods in Influence Maximization Research as well as Link Prediction in Graph Learning to determine influencers on social networks. Furthermore, we will employ techniques in Multi-Agent Systems and Reinforcement Learning to compute an optimal action plan in terms of how frequently each influencer should spread dietary information from our program to his/her peers on social networks. The novelty of our approach is the integration of influencer determination and action plan computation into a single pipeline, with our goal being to maximize the impact of our MyThali program in the entire network population. In fact, these AI techniques have been applied successfully for solving many other real-world problems. For example, methods in Influence Maximization Research have been applied successfully for raising awareness about HIV among homeless youths based on the youth social network .
Finally, results (based on anonymous data) from this project will be published at top AI conferences such as AAAI (Association for the Advancement of Artificial Intelligence) and IJCAI (International Joint Conference on Artificial Intelligence).
Also noteworthy is that this same project team has won the 2021 Google AI for Social Good award earlier this year and is using AI to improve the effectiveness of Arogya World’s text message-based diabetes prevention program, mDiabetes, in South India.
 Yadav et. al. Bridging the Gap Between Theory and Practice in Influence Maximization: Raising Awareness about HIV among Homeless Youth. In IJCAI 2018.
Scaling of impact to SDGs: Please detail how many citizens/communities and/or researchers/businesses this has had or can have a positive impact on, including particular groups where applicable and to what extent.
Arogya World promotes NCD prevention through healthy eating because:
• Unhealthy eating is the #1 cause of death – Global Burden of Disease Report 2019.
• 80% of heart disease and diabetes can be prevented by eating right, improving physical activity and avoiding tobacco – WHO.
• Transformation to healthy diets by 2050 will require substantial dietary shifts – >50% reduction in global consumption of unhealthy foods, and >100% increase in consumption of healthy foods – EAT Lancet Commission 2019.
• This is the Decade of Action on Nutrition (2016-2025)
We developed MyThali in picture form, following the US MyPlate approach, to show urban Indian women, who are often the decision-makers about food in the family, what to eat and in what quantities at each meal.
We promote balanced meals and portion control through social media (@my.thali on Instagram), by reaching food enthusiasts, and highlighting the voices of everyday people, influencers and chefs. We also promote eating right via traditional print, online and broadcast media. Our total reach (Dec 2019 – July 2021) has been >21 million people (>1.2 million via Instagram, >4 million via Foodies groups, and nearly 16 million through traditional media).
We believe we can help India meet SDG #3 – health and wellbeing for all at all ages. MyThali is an innovative tool that can help end malnutrition in all its forms. Eating right is the responsibility of our generation. We must eat right to leave the world a healthier place for our children and the next generation.
AI can augment the effectiveness of MyThali by helping us sharpen our focus on the most influential. This project will shine a light on the power of AI to improve lives leveraging modern day social networks in smart ways.
Scaling of AI solution: Please detail what proof of concept or implementations can you show now in terms of its efficacy and how the solution can be scaled to provide a global impact ad how realistic that scaling is.
We aim to reach 100 million by Dec 2023. That kind of uptake is possible because of modern day communication methods – traditional, online, and social media, online food delivery channels etc. We are already ¼ of the way there.
We are working on a cost-effective evaluation plan. We will field an eating habits questionnaire among various consumer groups, once in 6 months, and work with NielsenIQ to evaluate the responses. We expect this questionnaire to be rolled out from Oct 2021, timed to a planned MyThali pledge campaign. We will consult our PR firm partner, and our high powered MyThali Advisory committee made of India’s top nutrition experts, to help interpret the results. We expect to see an uptick in effectiveness once we begin to use AI. We expect AI to help 5% of the Indians ie 5 million to improve their eating habits by end 2023.
This effort is already a collaboration between civil society (Arogya World) bringing public health expertise, and academia (Dr Thanh from Univ of Oregon) bringing the needed academic expertise. If picked by UNESCO, we expect it to spur more collaborations globally between academia, industry, government and civil society to truly harness the power of AI for social good. We believe we can change the way India eats with MyThali and the use of AI. That will be an incredible case study for the use of AI on the global stage.
Regarding the scalability and sustainability of our AI solutions, the influence maximization methods that we use do not require or depend on any domain-specific characteristics, and thus can be generalized and applied to any social network domains, globally. We can simultaneously employ these methods in multiple global communities, allowing our program to be scaled up to large populations
Ethical aspect: Please detail the way the solution addresses any of the main ethical aspects, including trustworthiness, bias, gender issues, etc.
The project team pledges to meet all aspects of trust, ethics and integrity in the responsible use of social media to promote healthy eating. The MyThali project is designed to create interest in people to learn more about healthy eating and adopting a positive lifestyle. Only those consumers who are interested would sign up for various Calls to Action, Pledge campaigns and recipe contests on our social media channels. Or access our brochures, celebrity chef cookbooks, and customized meal plans on our website. Again, only interested consumers would sign up for our year-long healthy eating email campaign. In surveys we conduct, we are most interested in assessing eating habits – eg number of vegetables and fruits consumed, eating out behavior, avoidance of fried foods, etc. We will not collect medical data and will follow accepted practices of data security. Because we target AI methodologies among urban consumers in India, who are by and large tech savvy and participate in social networks actively, we do believe this project, by design, is equitable. By targeting women, this effort already has an in-built gender focus and aims to achieve inclusiveness in social media, a space in India where more (2 fold more) men participate.