Digital Twins: The Future of Personalized Nutrition and Health?

Digital twins are created by combining data from various sources, such as medical records, genetic information, and wearable devices [3]. These data are fed into a computer model, which creates a virtual representation of the patient’s body and its biological functions. The model can be used to simulate different scenarios to predict the outcome, such as how a particular drug will affect the patient, or how changes in lifestyle will impact their health [4].

One of the key advantages of digital twins is that they can be updated in real time, based on new data [4]. This means that doctors and researchers can monitor the patient’s condition and adjust their treatment plan accordingly. For example, if a patient’s blood glucose level starts to rise, the digital twin can alert the doctor or practitioner, who can then recommend changes to their diet, physical activity, or medication.

Recent reports have confirmed that chronic diseases such as cardiovascular disease and diabetes are large contributors of healthcare expenditure [5], yet technology that can provide accurate and real-time information into the status of an individual has sorely been lacking. Digital twins have the potential to revolutionize the way we approach healthcare by adopting a more preventative and proactive approach in a data-driven way [4].

The advent of big data and computational power has meant that the opportunity to provide real-time feedback, supported by a practitioner, is becoming a reality. Digital twin technology could predict the likelihood of a patient developing diabetes based on their genetics, lifestyle, and other factors [6]. This information could then be used to create a personalized prevention plan, which could include changes to their diet, exercise routine, and medication.

Personalized nutrition is an emerging field that aims to develop customized diets and supplements based on an individual’s unique needs, health goals, and preferences with the goal of improving health and nutritional outcomes [7]. Recent studies have demonstrated that individual responses to identical meals vary greatly in terms of post-meal glucose and triglyceride levels [8‒10]. These differences can be attributed to factors such as genetics, microbiome composition, meal composition, timing of meals, or even life stage [8‒10].

Digital twin technology has the potential to revolutionize the nutrition field by providing doctors and nutritionists with a more accurate picture of the patient’s nutritional needs and response in real time. By creating a virtual model of the patient’s body and its functions, digital twins can identify potential deficiencies or imbalances in their diet and recommend changes to improve their health using a collection of different data points such as diet, physical activity, BMI, age, and metabolites levels [6].

Wageningen University in the Netherlands recently completed a digital twin project in metabolic health [11, 12]. The aim of the project called “Me, my diet and I” was to develop a personalized digital twin that can predict changes to an individual’s blood values such as blood glucose and triglycerides to provide dietary advice through an app, with the goal to reduce cardiometabolic disease risk [11, 12]. The multidisciplinary team believes that a digital twin’s advantage lies in the fact that personalized dietary advice can be automated without the initial or immediate input of a healthcare professional such as a dietitian [12]. This approach could reduce the cost associated with acquiring the support of credentialled practitioners, while potentially increasing the level of self-efficacy of an individual [13].

The body of research relating to digital twins in healthcare is still very small, with only 88 papers published in 2023 [14]. One of the most promising applications of digital twins in healthcare is in the field of diabetes research. A recent study conducted at the University of Warwick (UK) sought to predict the progression of type 2 diabetes using digital twin technology [15]. The study used data from over 1,000 patients which included medical history, genetics, and lifestyle factors. The researchers created personalized digital twins for each patient, which were used to simulate the progression of their disease over time.

The results of the study were impressive. The digital twins were able to accurately predict the progression of diabetes in over 80% of cases, and they identified several key risk factors that were previously unknown [15].

In a separate retrospective study, digital twin technology was used to provide personalized nutrition advice via expert coaches [16]. The results demonstrated that daily precision nutrition guidance based on continuous glucose monitoring, food intake data, and machine learning algorithms can be beneficial for patients living with type 2 diabetes. At the end of the 3-month period, patients achieved a 1.9% decrease in HbA1C, a 6.1% reduction in body weight, as well as a 56.9% reduction in HOMA-IR. This demonstrates the potential effectiveness of using digital twin technology in diabetes.

Personalized nutrition crosses the spectrum from prevention to treatment. Malnutrition is a common problem in healthcare, with an estimated 20–50% of elderly patients admitted to hospital already being malnourished [17, 18]. Malnutrition can worsen health outcomes and modify the response to treatments, which is why early detection is crucial [19].

To date, many screening tools have been developed, such as Global Leadership Initiative on Malnutrition (GLIM), Malnutrition Universal Screening Tool (MUST), and Short Nutritional Assessment Questionnaire (SNAQ) to screen for malnutrition on admission and on the ward, yet many patients fail to be screened during their treatment journey especially on hospital discharge [19]. Nutritional support can increase survival after hospital discharge and needs to be personalized for every patient based on their medical history, nutritional status, and social determinants [19]. This may require prescription of specialized diets, nutritional supplements, or the instigation of enteral or parenteral nutrition.

A digital twin in this scenario could be used to simulate and track responses of a patient to prevent adverse outcomes as well as reduce risk of worsening malnutrition. A recent paper highlighted that data points such as weight, height, age, calf circumference, grip strength, and dietary intake information could be combined with hospital and electronic health records to provide a detailed picture of each patient [3]. As medical and nutritional treatment is instigated, the digital twin status automatically changes providing prediction of the patient status and predicted outcomes in real time [3]. This could have important implications for lowering hospital readmission rates as well as reducing health-related expenditure.

Based on the most recent survey by IFIC (2024), at least 52% of Americans follow some type of dietary pattern with “high protein” currently taking the lead for individuals looking to manage their weight [20]. However, it is well known that not everybody responds to dietary patterns in the same way. What if it was possible to take out the guesswork before even starting?

Researchers recently created an offline digital twin using different, non-connected, and complementary information about human metabolism into a single, quantitative, robust, and coherent picture [21]. They demonstrated the potential of a digital twin by simulating the impact that a variety of different diets are expected to have on key variables, such as mean plasma glucose, plasma insulin, and liver glycogen. Many of these data and predictions are subject-specific, while others are population-specific potentially, therefore ultimately impacting population health. The researchers believe that this type of offline digital twin could become useful, not only for personalized nutrition, but also to improve patient understanding, motivation, adherence, health, and behavioral outcomes [21].

Behavior change is notoriously complex and can take a long time [22]. Behavioral data could provide detailed insight into the daily choices of an individual with the goal of receiving real-time personalized feedback to drive behavior change through a digital twin.

Advanced digital technologies have integrated behavior change techniques (BCTs) into products to deliver behavior change at scale [23]. Despite numerous BCTs available, a recent meta-analysis demonstrated that a limited number of three BCTs are frequently used in nutrition apps; these include goal setting and planning, social support, and self-tracking [23]. This means that our understanding of how to modify behavior by leveraging digital is woefully inadequate and incomplete.

In addition, there are limited data on consumer acceptance of digital twins, or whether having a twin will drive meaningful behavior change [24]. While there is optimism for the future impact of technology on health and behavioral outcomes, the future of digital twin technology in behavior change is still unclear.

Digital twins, if implemented correctly, can reduce health inequalities that exist in healthcare systems according to a recent Ernst & Young publication [25]. As the pandemic brought to the forefront the racial and social inequities that persist in healthcare, health systems have an opportunity to use digital twin technology and predictive analytics on a population level, to identify barriers to care and help balance outcomes [25]. By stratifying the risk for different segments of the population, providers as well as payers can create digital community twins that help enable them to identify vulnerable populations and develop different tracks of outreach that reflect the community needs and prioritize resources accordingly [25].

Digital twins have enormous potential in healthcare and personalized nutrition as outlined above, yet several challenges need to be addressed. Besides the obvious high cost associated with developing a digital twin, one of the biggest challenges is data privacy and security. Digital twins rely on large amounts of personal data, which must be protected from unauthorized access or misuse [26].

The data required to develop a digital twin range from personal, lifestyle, genetic, and biological data. In order to ensure equity and inclusion, these data need to include relevant data from a specific population (e.g., American vs. European vs. Asian vs. Latin American). For example, BMI categories differ between European and Asian populations and specific risk alleles may vary by population [27]. In addition, environmental and cultural factors vary widely and these need to be considered and included when developing a digital twin [28].

In addition, research conducted by Wageningen University demonstrated that consumers trust the advice provided by a healthcare professional to be more reliable and trustworthy [11]. This means that future initiatives need to ensure that consumers understand the benefits and role of digital twins in relation to their healthcare.

Individual preferences for receiving dietary advice may differ, and a multidisciplinary approach must be adopted in the development of a digital twin to prevent unwanted, incompatible, and irrelevant personalized dietary advice being provided [12, 29]. As with other digital health technologies, their integration into workflows needs to be carefully considered on a case-by-case basis, with the sole purpose of benefiting the patient or users by optimizing health, or supporting behavior change without causing harm. A recent systematic review and meta-analysis have demonstrated the potential benefits of digital healthcare technologies in terms of practitioner competencies and decision-making [30]. Whether the personalized advice provided to individuals will be more personalized and result in lasting behavior change is yet to be confirmed.

Another challenge is the ethical implications of using digital twins for healthcare applications. For example, there are concerns about how digital twins could be used to discriminate against certain groups of patients, or how they could be used to deny insurance coverage based on predicted health risks [31]. It was only a decade ago when concerns were raised about whether knowing one’s genotype would lead to meaningful behavior change or whether this would result in a fatalistic attitude and demotivate individuals [32].

Finally, training of healthcare professionals will be necessary to ensure competency in interpreting data and recommendations from digital twins [33]. This will require a concerted approach to ensure equity across health systems.

Despite these challenges, the outlook for digital twins in healthcare and personalized nutrition is bright. It is estimated the global market for digital twins will grow from USD 10.1 billion in 2023 to USD 110.1 billion by 2028, at a CAGR of 61.3% [34].

Digital twin technology has already been implemented across numerous industries, and its potential is being evaluated in healthcare. As technology continues to advance, we can expect to see more sophisticated and accurate models, which will help doctors and researchers develop more effective nutrition treatments and disease prevention strategies. However, in order for digital twin technology to succeed, many challenges will need to be addressed such as cost, extensive education of practitioners and consumers across the healthcare spectrum, and trust. With the right safeguards in place, digital twin technology has the potential to revolutionize the way we approach healthcare and nutrition to improve the lives of millions of people in an equitable way.

The author is an industry consultant providing consultancy and speaking services. This editorial was not supported by any sponsor or funder. No professional relationships have influenced the writing of this editorial.

M.A. conceptualized and wrote the editorial.