January AI Looks to Revolutionize Personalized Diabetes Care
Noosheen Hashemi is the founder and CEO of January AI, a precision health company designed to encourage small lifestyle changes and habit-building that add up to a significant improvement in one’s health, with an initial focus on type 2 diabetes. Following the presentation of impressive data at ADA 2020, Hashemi spoke with Beyond Type 2 about the company and its future.
Beyond Type 2: Can you tell us a bit about your background? How did you come to work in the health + diabetes sector? Can you tell us about the founding of January AI?
Noosheen Hashemi: I worked at Oracle for 10 years, then in the mid ’90s I went to a start-up for a year, and then began running our family office, investing in markets and nonprofits. Two kids, and two foundations later, in 2013, I decided to go back to the private sector full-time and started active-investing in startups. The next logical step was to take on the ultimate challenge and start a company from a clean sheet of paper. I happened to attend a conference that my friend and MIT economist, Andrew Lo, was convening on how to supercharge investing in medical research. Reverse engineering five billion years of evolution is nontrivial and we still know so little. Because of the war Nixon declared on cancer four decades ago, we now have a few therapies and solutions for cancer. We have very little for neurological disease and have nothing for the ultimate disease, aging. Then I attended a Stanford conference on the societal impact of AI and one session was on health and I was hooked. There, FeiFei Li recommended the upcoming Machine Learning in Healthcare conference. I booked my ticket to LA that night and attended two weeks later. A light bulb went off in my head when I learned that ML would help us fill in for missing variables in research. Boom! We could accelerate what we know about health and move the needle on human suffering, on cost, on happiness. At MedicineX, I became obsessed with the patient’s voice and the necessity of healthcare to undergo consumerization and put the person, the human, at the center of it. My research then led me to whole-person health, systems biology, and multiomics, which led me to my co-founder, Mike Snyder.
Mike has type 2 diabetes and has conducted longitudinal multiomic studies with continuous glucose monitors. He was excited that for the first time, we were thinking of wearables not just for fitness but for health. We dreamed about helping people who have pre-prediabetes and prediabetes see inside their bodies, dial their lifestyles and not let their blood glucose dysregulation tip into diabetes. Given the urgency of diabetes, we eventually decided to address the spectrum of metabolic syndrome. We are fundamentally committed to helping people advocate for and take charge of their own health. We believe in giving people access to their own health data, the ability to combine it across devices and apps and curated insights to act on. Agency is January’s currency.
Tell us more about whole-person health.
Too often we reduce people to single markers like their cholesterol, weight, or A1c. But people and their health are more nuanced than that. Ideally, you want to know everything about a person to see where they are. Their genetics, foods they ate growing up, their microbiome, the kind of work they do, the amount of toxicity in their environment, their level of stress at home and at work and so on. Most importantly, you want to know about their daily habits: what they eat, how much they move and sleep. We wanted to develop the most affordable and the most personalized solution for people with diabetes so we decided to start with a few simple pieces of data: food consumption, heart rate and blood sugar. With this information, we have been able to glean deep insights into metabolic health and believe that we can transform how we care for diabetes. The biggest solution out there today is using 1972 technology: glucometers and strips which are painful to use and give you only partial data. They manage a single marker, A1c which is something you test for every 90 days. Most solutions provide generic advice like don’t eat sugar and eat vegetables and work out which is not effective. No wonder why the top solutions out there serve fewer than one million people.
We intend to integrate with lab and EHR data, microbiome data, blood pressure and so on. We hope to help people build a unified picture of their health and get the synthesis and analysis that they need to make use of the data in a meaningful way.
What was the Sugar Challenge Study? What did the study show?
It was a 10-day observational study, and we had 23,000 people apply. We finished with 1022 healthy volunteers: People with pre-diabetes, and people with diabetes. At least 250 of them had type 2 diabetes.
The study really makes the case for personalization. Because, people are just vastly different, and their diabetes is also different. So, if you eat the same meal, let’s say, nine days in a row, there may be a different glycemic response to those identical meals based on activity, sleep, stress, fiber intake, etc.
And through the study, we found these incredible differences amongst people, and how their diabetes manifests. Machine learning can pick up small differences and we were able to discern how we can specifically help people with hyper-personalized insights and recommendations. We are also able to predict their future glycemic responses to food and activity, even when they are no longer wearing a continuous glucose monitor (CGM). The poster at ADA reflects what we call CGP—Continuous Glucose Prediction. We predicted blood glucose for 33 hours for people with a high level of accuracy. This study essentially was the first step. And we’ll do other studies. We will add microbiome for a study next. And, we have a huge interest in the microbiome.
Can you tell me about the January AI app? How does it work?
There are many cool parts of the app, but one of the coolest is that it seamlessly and effortlessly overlays your food and CGM readouts and your heart rate and puts them together. When you go to talk to a coach or a doctor, you won’t have to walk in with 90 pages of your AGP [Ambulatory Glucose Profile] report and try to explain what you were doing when you had a big spike. January shows your biggest spikes and what caused them, what foods you could substitute, and or what amount of walking you could undertake to bring your blood sugar into a healthy range. The primary goal of the app is to help you maximize your Time in Range.
Fiber modulates the microbiome and is important for flattening blood glucose curves. The app tracks how much fiber you’re eating and encourages you to eat more. Intermittent fasting has been shown to be beneficial for insulin resistance. The app shows your fasting period when you first start using it and it nudges you to increase it over time. We have a 30-day clinical program, Season of Me, that walks you through various experiments and you can see the impact of various lifestyle actions on your blood sugar, not in general, but specifically, for you.
Would the app work for people who are on insulin? With insulin-dependent Type 2 diabetes?
We do plan in the future to work with insulin as one input, but it’s not our focus right now. Most of the insulin prediction models use simple carb to insulin ratios, and a lot of people already use those. Our focus is on minimizing glycemic load and increasing fiber as key mechanisms for managing blood sugar and building a healthy microbiome, and on encouraging the behavior changes that lead to lasting health improvements. Today, we are focused on whole-person health rather than A1c and medication management only. While we don’t see ourselves focused on insulin-dependent type 2 diabetes in the immediate future, ultimately, we want to serve this population as well.
Is there anything else you can tell us about January AI?
There seems to be a lot of stasis in the way that we think of diabetes and pre-diabetes, as well as diabetes management. We are at a watershed moment in embracing technology in healthcare, and I hope it isn’t wasted.
I hope that providers, payers, and employers keep an open mind about adopting new ways of serving their constituents. DPP was created before Dexcom, FreeStyle Libre, smartphones, fitness trackers, smartwatches. Why take 16 weeks of the generic curriculum to learn how your body works when you can do that in two weeks? When you can change habits in three?
We have the tools to help people see inside their bodies, get to know their own diabetes, and figure out which interventions have the greatest impact on their health. Fasting will come easily to some, it will be harder for others. Some people will incorporate more fiber into their diets, some will need to take supplements. Some folks want to know the price of eating something in minutes of walking and will act on that guidance and some won’t. Welcome to the world of hyper-personalization. No two bodies are the same, why should advice and treatments be.
We have the tools here today to help people spend less time managing their diabetes, bring more joy to eating, give specific recommendations about activity and let people gain agency over their own care, be the CEO of their own healthcare, as a wise person recently said.
