January AI adds next-level digital twin app for food companies
The company initially created its AI app, January, that can predict glucose response and how consumers will react to foods before consuming those items.
Now, the company is working with Mars, the multinational provider of food, pet food, confectionary and other products, and its Mars Advanced Research Institute (MARI) to leverage these generative AI tools. They want to determine how customers will respond metabolically to a variety of foods and formulations across subpopulations.
Noosheen Hashemi, CEO of January AI, said that Mars is quiet about how its R&D works, but that her company is looking forward to doing more work with companies that voluntarily choose to create healthier foods.
“I think Mars wants to be supportive of good health and be conscious,” Hashemi said. “I think Mars is trying to be a good company, and it's trying to say, ‘hey, we care about the foods that we put out there.’”
She added that partnering with the company is a “milestone."
Darren Logan, vice president of research and science discovery at Mars, said in a statement that the company was excited to work with January AI and that "this type of innovative and science-driven joint program enables Mars to learn more about the role of AI technologies in our future research and development programs.”
Last year, January AI announced a partnership with Nestle to also utilize its digital twin technology.
As a generative AI company, January AI's technology is backed by peer-reviewed science, using millions of data points and, as of earlier this year, has the largest Glycemic Index (GI) and Glycemic Load (GL) database of 32 million foods.
Harnessing digital twins
The basis for the digital twin technology lies in the January consumer app. Whether one consumes a supplement or a piece of cake, January tracks it all in real time, accounting for blood sugar without having to wear a continuous glucose monitor (CGM). Users take photos of a meal or food with their smartphones and gain insights into ingredients, glucose and meal alternatives.
To gain data for the digital twin, January AI then focused on a non-invasive data collection process, starting with wearables. Researchers did include information from users who wore CGMs but for a limited time. Instead of wearing the CGM for 12 months, a user could wear it once, and the AI model would still have enough data to create health predictions. Researchers would recommend wearing the CGM more frequently if a major life change, such as pregnancy or significant weight gain, occurred.
No company has created a digital twin of an individual’s entire human body. Instead, they have developed replicas of specific processes of the body, which is what January AI has done with users’ metabolic systems—kind of like building a mathematical model, Hashemi said.
“Our system doesn’t predict your brain waves, or things like that,” she added. “It’s not an entire twin of you, but it's a digital twin of your metabolic system, which means we have figured out the inputs and outputs.”
January AI researchers collect a person’s glucose and smartwatch readings, detailed food log data and information from other user-reported questions. Over a period of days, the company can observe everything a person eats, how they respond to food during workouts or activity, and any effects during sleep to determine baseline data.
“We are able to predict from then on what happens to you [physiologically],” Hashemi said.
As for Mars, it was using the January AI digital twin to input some of the foods the company has on the market and in development. Twins included those with diabetes, prediabetes and healthy people and were 'fed' recipes users might eat and on an empty stomach, which is how glycemic response is normally measured.
Hashemi said companies are given an analysis and it helps them figure out how to modify the recipes to create foods that are supportive from a health standpoint.
The birth of January AI
Hashemi may not be the most likely person to enter the personalized nutrition or precision health space. She was previously a vice president at Oracle, has been an associate on the Council of Foreign Relations and was a board member of the New America Foundation. Her career trajectory changed in 2015 when she decided to start a company and jointly became interested in health, partly because she was dealing with her own medical concerns as well as those of her parents.
“I became very interested in prevention,” Hashemi said. “What I later learned was that it was called precision health, which is upstream from precision medicine. You’re able to look at a bunch of data and see if you can predict that someone might get sick and how they might get sick.”
It also led her to the field of multi-omics, which examines health from different angles. This could mean exploring the outcome of blood tests but also a person’s genomics and microbiome through wearable data to understand present and future health. Her interest in multi-omics led her to partner with Michael Snyder, the co-founder of January AI, who had been studying multi-omics work at Stanford University and is director of the school’s Center for Genomics and Personalized Medicine. Snyder also has type 2 diabetes and had worn a CGM for years. From a commercial standpoint, he was interested in help keep people with prediabetes from progressing to diabetes.
The company focuses on blood glucose because it is a key measure of health. The metabolic system involves the hormone insulin that allows a cell to take up glucose.
“If you're not producing enough insulin or you don't produce it at all, or you don't produce it fast enough, you may end up with a situation where you have too much glucose in your in your bloodstream...So you're going to not feel good and feel lethargic, but not necessarily know you are in a pre-diabetes condition,” Hashemi said.
She also worries that the food industry is not responding to this public health concern and not taking AI as seriously as the drug industry is.
“I don't see them taking this on with gumption,” she said. “Health is not the only reason why you should use AI to develop healthier foods. You can also develop better tasting foods, and you might also be looking at different ways of formulating foods that you haven't thought about before.”