The NEW Science of Nutrition

The principle of a digital twin is simple and self-explanatory. It is a digital replica of an individual's chemical complexity, with the metabolic pathways living and interacting inside the computer. In practice, however, it is a stripped-down version of a real person, beginning as a simple digital molecular network and growing in complexity as more and more data about the individual are fed into the computer. With this, some engineers and scientists believe, doctors can reverse metabolic diseases such as diabetes.

The idea drew the interest of Jahangir Mohammed and M.A. Maluk Mohamed. The two engineers-cum-entrepreneurs had met at the United Economic Forum Trade Summit in Chennai in 2017. Mohammed was a serial entrepreneur in Silicon Valley who had just sold his company, Jasper Technologies, to Cisco for $1.6 billion. Mohamed was the Chairman of the M.A.M. Group of colleges in Tamil Nadu. Mohammed was looking for opportunities in healthcare after selling his company, and Mohamed was looking to develop healthcare technology after a difficult family experience when his young son's liver failed due to unknown reasons.

The two entrepreneurs joined hands with Terry Poon, who was Vice-President of Engineering at Jasper, to set up Twin Health in 2018, with offices in Chennai and Silicon Valley. Over six years, they developed a digital twin technology that could be used by experienced doctors to provide advice to treat or reverse metabolic diseases. "As we dug deeper," says Mohamed, "we found that there cannot be a common solution for every human because the biology of every human is unique." Human metabolism is complex and dynamic, and its defects are the root causes of several diseases. Metabolism varies from individual to individual. It also varies in the same individual as they age.

The concept of a digital twin was anticipated by space engineers and authors through the 1970s and 1990s. The first concrete idea was proposed in 2002 by NASA engineer John Vickers and product life-cycle management specialist Michael Grieves. When Twin Health was set up, the idea had become popular among manufacturing companies; organisations were already developing digital twin technology for healthcare. The French company Dassault Systèmes developed a digital living heart in 2014, a product that had been improved over the years and used to understand cardiac functions. Since 2018, global biopharma company Takeda Pharmaceutical has been using a digital twin — developed by PwC Consulting — for testing drugs.

Twin Health was specifically interested in understanding differences in the habits and metabolism of different individuals and then using this knowledge to reverse diabetes and connected issues. They saw diabetes as a problem of biology as well as of behaviour. Insulin and drugs such as metformin controlled blood sugar, but did not address the root problem. Lowering blood sugar in people with diabetes reduced damage to vital organs, but it was not a cure. Dietary advice also focused on macronutrients such as carbohydrates, proteins, and fats. "The truth is macronutrients are just a source of fuel to the body," says Mohamed. "You need to look at the micronutrients, too, to be able to provide effective care."

A whole-body digital twin measures an individual's metabolism using a vast spectrum of input data. This includes data from devices such as glucose monitors, smart watches or activity trackers, and a body composition scale. The inputs include causal data, relating to diseases and medications, and data from blood tests, food intake, and physical activity. A total of 57 input parameters are put together to make a digital twin. The first set of data points creates a rudimentary version. But the twin evolves as data from daily activities are fed in over time. A user gets a metabolic score, which is an indicator of the health of the individual's metabolism and, hence, the progress in improving overall health.

In a 2021 study in Clinical Diabetes and Endocrinology, researchers from Twin Health assessed the digital twins' ability to reverse type 2 diabetes (bit.ly/twinhealth_T2D). It was a retrospective study that looked at the impact of using Twin Health over 90 days. At the beginning of the study, most patients were on two diabetes pills. After 90 days, about 18% still needed them, while nearly half had stopped all medications and yet had healthy blood sugar levels. In a 2024 study (bit.ly/twinhealth_HT) in JACC: Advances, the researchers compared standard care with the digital twin in reducing symptoms of hypertension in patients with diabetes. Those on the digital twin were significantly better after one year; 68.2% of the digital twin group remained off blood pressure medications compared to none in the control group.

The use of digital twins has been growing in healthcare in areas such as drug development, preventive healthcare, diagnoses and treatment optimisation. It promises to significantly impact the treatment of lifestyle diseases such as diabetes and obesity. It also promises to impact what and how people eat. "People have realised that the fundamental way to bring about a change in the biology of an individual is through nutrition," says Mohamed. "Unmonitored dieting is evil for a body with disrupted metabolism."

Nutrition researchers hold that diet is a lifelong exposure. It impacts health all through a person's life, especially when it comes to lifestyle diseases. In the last few decades, the prevalence of lifestyle diseases has been increasing significantly across different populations. Scientists know that diet plays an important role in these diseases but don't fully understand how specific food choices lead to illnesses. This is because multiple aspects of diet matter, and so there is no direct and measurable link between diet and disease. Researchers are now looking at ways of connecting the dots between diet and health and using data from genetics, the microbiome, and more accurate measurements of food intake. "In the last 5-10 years, the most notable shift in research includes increased attention on the need for more personalised, multidisciplinary, and sustainable approaches to solving the long-intractable and persistent problem of malnutrition," says Saurabh Mehta, the Janet and Gordon Lankton Professor in the Division of Nutritional Sciences at Cornell University. "That one-size-fits-all approaches will not solve all our current challenges in nutrition is now recognised not only across research but also the donor, programmatic, and funding communities."

South Asians are 23% of the world population and are at twice as much risk for getting cardiovascular disease and four times as much risk for diabetes, compared to Europeans. However, South Asians form only 1.3% of the participants in genetic studies on disease risks. In 2023, doctors and scientists from major medical institutions in the U.S. — such as Harvard, Stanford, Yale, and MIT — came together to propose a study to understand the genetic basis of this increased risk and to generate improved guidelines for them. The OurHealth study collects genetic, lifestyle and clinical data from South Asians in the U.S. to understand how genetics and lifestyle can combine to increase the risk of disease. "Some of the guidelines that you have for managing obesity or having an ideal body weight don't really apply (to them) because the cut-offs are so different," says Shilpa Bhupathiraju, who is associated with the OurHealth study and is an Assistant Professor in the Department of Nutrition, the Harvard T.H. Chan School of Public Health. "In our work, we have to ask: Who do your results apply to? Does it apply to everyone or only a subset?"

For example, several hundred genes are known to affect obesity. In 2012, in one of the first studies (bit.ly/FTO_gene) on the impact of genetics on diet, researchers at Harvard and other institutions tested the effect of the fat mass and obesity-associated (FTO) gene on weight loss over two years of treatment. There are different variants of this gene, and one was found to be associated with greater weight loss. Other genes, such as the insulin receptor substrate 1 (IRS1) and glucose-dependent insulinotropic polypeptide receptor (GIPR) gene, also appeared to impact weight loss (bit.ly/weightloss_genes). Such studies provided clues as to why individuals respond to specific foods and diets in different ways and how diets affect disease.

Vimal Karani, Professor of Nutrigenetics and Nutrigenomics at the University of Reading, U.K., has studied the FTO gene in different populations. The gene exists in three different forms: high, moderate, and low genetic risk for obesity. "If you carry the high genetic risk form of the FTO gene, then you have an increased risk of becoming obese," says Karani. According to him, the prevalence of the high-risk form is 25-30% in Europeans, and nearly 40-45% in the Indian population (bit.ly/FTO-Indian). A 2024 review by Karani and colleagues in Nutrition Reviews (bit.ly/gene-diet_Indian) summarising the contemporary evidence on the impact of diet and other lifestyle factors on metabolism notes that the variants of FTO and the TCF7L2 (gene) have been found to have significant interactions with diet, including carbohydrates, protein, and fibre, and influence the risk of diabetes in the South Asian population.

In individuals with high-risk genetic variants, diet is a modifiable factor that can reduce the genetic risks for diseases. In a recent pilot study (go.nature.com/3XP3doU), researchers compared general dietary advice with gene-based personalised advice in patients with impaired glucose regulation but who are not yet diabetic. At the end of the trial period of 26 weeks, there was a significant reduction in fasting plasma glucose in people who received DNA-based advice compared to those who got general advice.

Mapmygenome, a Hyderabad-based company, offers nutrition advice based on gene data culled from large genome databases. "These databases offer a wealth of data on genetic variant frequencies across populations," says Anu Acharya, Founder and CEO of Mapmygenome. Apart from the FTO gene, other genes include the MTHFR (folate metabolism and methylation efficiency), TCF7L2 (impacts carbohydrate metabolism and diabetes risk), CYP1A2 (caffeine metabolism and tolerance), APOA2 (influences saturated fat sensitivity), and GC (Vitamin D binding protein). The company says that individuals with a high sensitivity to caffeine, based on CYP1A2 variants, have adjusted their caffeine intake to reduce issues such as anxiety and disrupted sleep. Those with MTHFR polymorphisms have benefited from targeted vitamin supplementation, such as an increased intake of methylated folate, to address fatigue and optimise energy levels.

At the Bengaluru-based nutrition clinic Qua Nutrition, Ryan Fernando advises an elite clientele based on genetic and microbiome (gut bacteria) tests. Although most of the microbiome and genetic data on nutrition come from the Western population, Fernando still sees value in using the inputs — till data on the Indian population are made available.

He gives the example of the HLA-DQ gene related to gluten sensitivity. Variants of this gene, HLA-DQ2 and HLA-DQ8, are associated with gluten sensitivity and celiac disease in the Western population. Genetic tests look at this gene to determine an individual's likelihood of getting gluten sensitivity. "When I get a positive on that, I go back to my client and tell them to avoid gluten for two or three months," says Fernando. "But before that, I take clinical observations from my client and/or blood testing data. People will listen to their genetic testing more than to Ryan Fernando."

Some researchers, however, feel that more population-specific data would be needed and thus caution against using such testing for dietary advice. "We don't have the evidence for the Indian population. That's why these kinds of diets could be disastrous," says Karani. He cites the example of obesity; hundreds of genes have been associated with it. "Some genes are shown to be associated with obesity in one population, but not in other populations. So we don't know whether it's a false positive finding or a true negative finding."

A similar trend of diet personalisation is also becoming common when it comes to the gut microbiome.

The human gut contains about one trillion bacteria, which consume and produce molecules that influence health. Scientists know that certain bacteria are harmful, and certain others beneficial. They also know that bacterial diversity promotes health. However, scientists have yet to figure out the precise relationship between diet and specific kinds of bacteria and their relationship to health. "If we lack the enzymes to break down certain foods ourselves, some microbes can do it for us," says Lindsay Hall, Chair of Microbiome Research at the University of Birmingham in the U.K.

For example, individuals genetically predisposed to lactose intolerance may not experience digestive problems if they possess higher levels of Bifidobacterium, a microbe that thrives on lactose. Because every person has a unique microbiome, personalised nutrition is a promising but challenging goal. And Hall's research on early-life nutrition shows the benefits of dietary interventions based on the microbiome. Diet in early life is primarily breast milk or formula. Breast milk contains special sugars called human milk oligosaccharides that only certain gut bacteria, such as Bifidobacterium, can digest. These bacteria are crucial for the healthy development of infants because they break down oligosaccharides into nutrients that the baby can use.

In a study published in Cell Reports Medicine (bit.ly/bifidobacterium) in 2020, Hall explains that they added a few drops of Bifidobacterium to the breast milk from the mother or a donor and fed it to premature babies, who often have low levels of this essential microbe and are at a higher risk for infections. This led to a healthier microbial balance, reduced harmful bacteria, and a decrease of over 50% in the risk of a severe infection called necrotizing enterocolitis.

In her most recent research, published in BMC Genomics in 2024 (bit.ly/hall-bacteria), she compared the gut bacteria in infants in Zimbabwe to those from the U.K. and other regions. She found that all infants have beneficial bacteria but there were differences in how these bacteria digested breast milk. She also identified the local bacterial strain that could help Zimbabwean infants digest breast milk.

Adults have a more mature microbiome than infants. It consists of several kinds of bacteria capable of digesting various types of molecules. So, tailoring nutrition based on the microbiome is much more challenging for adults. Besides nutrition, factors such as antibiotic use, existing health conditions, and lifestyle changes further complicate the interaction between diet and the microbiome.

The complex interactions between diet and the microbiome are shown in a study by researchers at Arizona State University and other institutions, published in Nature Communications in 2023 (go.nature.com/3zCGTqQ). People who ate a diet specifically promoting healthy gut bacteria lost about 116 calories a day by excreting more undigested food and by-products from the bacteria in their guts. However, some people on the same diet lost more weight than others because their gut microbiome could ferment fibres in the diet. "You'll find that some individuals may lose some weight (with certain diets), and some individuals don't respond at all," says Hall.

The reason could be the difference in gut microbes that could break down specific diet components such as fibre. Resistant starch, a type of starch found in plantains and legumes that doesn't get digested in the intestine and reaches the colon intact like fibre, has been shown to enhance the abundance of beneficial bacteria such as Ruminococcus bromii. Gut bacteria are also known to affect the progress of diabetes. Before a person develops diabetes, they go through different stages, Karani says. These stages may span 3-5 years. Changes in diet alone may not be able to slow down this progress. The microbiome is known to be equally important.

In recent years, scientists have added another element that interacts with the diet and influences health: fat cells.

"If fat cells work fine, they are a perfectly good friend to you. But a lot of diseases are invited if your fat cell is not behaving as it should," says Babukrishna Maniyadath, a postdoctoral candidate at the University of Southern Denmark, who studies the types, distribution, and quantity of fat cells in different parts of the body.

Over the past 15 years, the understanding of obesity and its associated diseases has evolved significantly, revealing that fat does not always equate to poor health. One key difference between healthy and unhealthy obesity is where fat cells are present in the body. "A person who has butt fat or thigh fat is not as unhealthy as a person who has a big tummy or a big belly," explains Maniyadath. According to a Nature Reviews Endocrinology study (go.nature.com/3XVwLl0) in 2024, while individuals with healthy obesity may face a higher risk of heart problems, their overall mortality risk is comparable to that of people with average weight and significantly lower than those with unhealthy obesity. Currently, Maniyadath's research focuses on understanding this complex relationship between fat cells and diet. She is studying fat stem cells in mice — progenitors of fat cells — by looking at the impact of high-fat diets on their behaviour and development.

Four distinct stem cells in mice can develop into healthy fat cells. Preliminary studies on human fat cells, including samples from patients undergoing bariatric surgery, have revealed a similar diversity of stem cells in the fat tissue, hinting that this complexity might be a universal feature. Maniyadath is conducting a detailed study that includes mice on a high-fat diet for varying lengths of time, ranging from one week to 18 weeks. These periods represent different stages of obesity: the early stage (one week), where the mice are just starting on the high-fat diet; the mid-term stage (four weeks), where the mice are beginning to gain weight; and the long-term stage (18 weeks) where the mice have become significantly obese.

Unpublished preliminary findings suggest that diet does have an impact on stem cells. The diversity of stem cells remains intact after just one week on a high-fat diet, though their proportions begin to shift. As obesity progresses, this diversity can decrease, potentially leading to a loss of normal cell functions, a crucial feature of metabolic diseases. The study aims to identify the types of stem cells that lead to healthy fat cells, which will allow scientists to predict how individuals will respond to different diets based on their stem cell profiles. Such research is in its early stages but hints at essential connections between diet, stem cell behaviour, and obesity.

The goal of nutrition science has always been to optimise diet for improved health. The traditional way to do that has been to look for deficiencies and supplement them from the outside. Scientists know that this strategy is no longer enough. A variety of datasets from genetics, metabolomics and microbiome now have the potential to influence dietary advice. These datasets can help adapt the universal dietary advice to individuals and make it more relevant for them.

See also:

Health on your plate

When loss is gain