Scientists at the Weizmann Institute have built an AI-powered “digital twin” that uses the world’s most detailed health database to forecast future diseases. The model can reveal risks years in advance, test preventive strategies, and even chart a personalized health trajectory.
The model is powered by what is considered the world’s most comprehensive human health database, collected from more than 30,000 volunteers worldwide. The goal is to detect diseases before they occur, delay their onset, or, in some cases, prevent them entirely.
For individuals, this could mean the ability to run simulations of different treatments and lifestyle choices. What if a change in diet could reduce your risk of diabetes by half? What if a particular therapy proved to be the most effective for your unique physiology? The model aims to give patients and doctors answers to such questions with unprecedented precision.
A peer-reviewed study on this project, published in Nature Medicine, underscores the significance of this breakthrough for preventive care and the entire field of precision medicine.
25 years of meticulously collected data
Behind this digital revolution lies a massive scientific endeavor known as the Human Phenotype Project, launched by the Weizmann Institute in 2018. But the groundwork stretches back even further. Over the past 25 years, participants from across the globe have undergone detailed medical examinations every two years, creating a dataset unlike anything seen before.
These exams are comprehensive, covering 17 different body systems and generating a treasure trove of information. Volunteers undergo ultrasound scans, bone density testing, gene sequencing, and cellular protein analysis. Their microbiomes are studied through samples from the gut, mouth, and even the vagina. Researchers track sleep patterns, glucose levels, dietary intake, and even voice recordings – every detail feeding into a gigantic portrait of human health. The result is more than 260 billion data points, forming the most detailed health database in existence today.
For decades, individuals and doctors have struggled with the same problem: making health decisions without being able to predict their long-term consequences. Should you eat less sugar? Exercise more? Switch medication? Without concrete evidence, motivation often fades. But what if you could see today that your current lifestyle would put you on the path to diabetes in ten years? That kind of clarity, scientists believe, could turn vague advice into powerful motivation.
What is the biological age of the liver?
One of the first practical applications of this dataset has been the calculation of biological age for individual organs and body systems. The AI model compares physiological changes in 17 body systems against expected values for a person’s chronological age, gender, and body mass index. Any deviation provides insight into how fast or slow a system is aging.
For instance, monitoring blood glucose has revealed typical growth rates for different age groups and sexes. When the model detects abnormal patterns, it can flag prediabetes years before traditional medical tests would. In fact, researchers estimate that up to 40% of individuals classified as healthy under current standards may actually show signs of prediabetes.
The data has also uncovered intriguing gender differences. Men tend to age biologically at a steady pace, whereas women experience a sharp acceleration around the time of menopause. Researchers found that bone density decline correlates more closely with time since menopause than with chronological age. These insights could allow for far more precise timing of hormone treatments and other preventive interventions.
The concept of assigning a “biological age” to specific organs, such as the liver and the heart, opens up new possibilities. If your liver is aging faster than the rest of your body, the model can highlight this as an area of concern, giving you and your physician the opportunity to act before disease develops.
Life trajectory will show potential health risks
The project's long-term vision is even more ambitious: to create unified digital twins for each participant. By integrating every data point – genetic, metabolic, behavioral, and environmental – the AI model can simulate a person’s “health trajectory.” In other words, it can chart a personalized roadmap of future health, complete with risk factors, potential diseases, and recommendations for prevention.
Already, scientists are using digital twins to test how different interventions might change outcomes. Which diet is best suited for you? Which medication is likely to be the most effective with the fewest side effects? The twin can run these simulations before any real-world decision is made.
The Weizmann team is also working on making this information accessible. Plans are underway to develop an application that will enable study participants to track their health data and simulations in real-time. In the future, people may be able to log in and see how today’s decisions, whether a fast-food meal or a workout, might ripple through their personal health trajectory.
If successful, this technology could transform medicine. Instead of reacting to illness after it strikes, individuals and healthcare providers would be empowered to anticipate and prevent it. For now, the digital twin remains a research project. But as AI models continue to refine predictions, the idea of personalized preventive healthcare is edging closer to reality.
The next challenge will be to translate these scientific advances into clinical practice. Ensuring accuracy, safeguarding privacy, and integrating digital twins into everyday healthcare will determine the extent to which this approach can be transformative. If achieved, it could provide physicians and patients with a powerful new framework for decision-making, shaping the practice of medicine for decades to come.
Yet, this vision relies on the quality of available data: while the Human Phenotype Project has collected uniquely detailed records, much health information worldwide remains uncaptured, of low quality, or not yet digitized.
During the ICT&health World Conference 2026 this and other key themes shaping the future of healthcare will be explored in keynotes, presentations and several workshops. For more information, you can visit the ICT&health World Conference website.
