{"slug": "your-medical-history-may-soon-tell-ai-what-happens-next", "title": "Your Medical History May Soon Tell AI What Happens Next", "summary": "Researchers at Harvard Medical School, Dana-Farber Cancer Institute, and Mass General Brigham have developed an AI tool called Aladynoulli that estimates a person's risk of developing 348 diseases by analyzing medical history and genetic risk, trained on data from more than 683,000 people with up to 52 years of follow-up. Unlike traditional risk calculators that focus on one disease at a time, Aladynoulli assesses patients holistically, identifying 20 biological signatures that link health trends to multiple conditions. The tool aims to improve disease prediction so doctors and patients can take preventive action.", "body_md": "TL;DR — Key Takeaways\n\n- Aladynoulli uses medical history and genetic risk to estimate a person’s likelihood of developing 348 diseases.\n- The AI analyzes patients holistically rather than relying on separate risk calculators for individual conditions.\n- Researchers trained the model on data from more than 683,000 people, covering as much as 52 years of follow-up.\n\nA doctor sees a patient with high blood pressure, elevated cholesterol and a family history of heart disease. Another patient walks into the same office with the same diagnosis. On paper, they may look remarkably similar.\n\nBut they aren’t.\n\nOne may be on a path toward heart disease. The other may be at greater risk for diabetes, cancer or a condition neither patient nor doctor is yet thinking about. The clues may already be buried in years of medical records, scattered across different specialists, diagnoses and test results, along with genetic information that is difficult for a human to interpret all at once.\n\nA new AI system is designed to find those connections.\n\nResearchers at Harvard Medical School, Dana-Farber Cancer Institute and Mass General Brigham have developed an AI tool called Aladynoulli that can estimate a person’s risk of developing 348 different diseases by examining their medical history and genetic risk. Unlike traditional risk calculators, which generally focus on one disease at a time, the system attempts to understand the patient as a whole.\n\nThe name Aladynoulli is an amalgamation of Age-dependent, Longitudinal, and All-inclusive Dynamics and Bernoulli distribution.\n\n“People are thinking about what their health is going to look like over the next few years, especially with increasing intervention options,” said Alexander Gusev, an associate professor of medicine at Dana-Farber and one of the study’s senior authors. “This tool offers a path toward improving the prediction of future diseases so doctors and patients can take action to try to prevent them.”\n\nThat idea, getting ahead of disease rather than waiting for it to appear, is where the technology could eventually change medicine.\n\nFor decades, doctors have relied on relatively straightforward ways to estimate risk. A cardiovascular assessment might consider blood pressure, cholesterol, blood sugar, smoking history, age and body mass index. A breast cancer assessment might use age, family history and other established risk factors. Those calculations can be valuable, but they are usually designed around a particular disease.\n\nResearchers trained Aladynoulli using data from more than 683,000 people in three major research databases: the UK Biobank, the National Institutes of Health’s All of Us Research Program and the Mass General Brigham Biobank. The records represented as much as 52 years of follow-up and included information on 348 diseases.\n\nThe model looks for patterns. The researchers first identified 20 biological “signatures,” or combinations of health trends that tend to precede particular diseases. A history of high cholesterol, for instance, can contribute to a signature associated with cardiovascular disease. Genetic variations can add another layer of information. Because the signatures overlap, the same biological pattern can be relevant to several diseases.\n\nAI then learns how those patterns interact with a person’s medical history and how diseases tend to occur together, like looking at individual pieces of a puzzle and recognizing the picture they form together.\n\n“There is a lot of useful information in a medical record, both over time and across different disease areas,” said Giovanni Parmigiani, a Dana-Farber researcher and professor of biostatistics at the Harvard T.H. Chan School of Public Health. “That data would be difficult for a human to process in their head but tractable for a machine-learning model.”\n\nThe model also changes as the patient changes. A medical record isn’t static, and neither is Aladynoulli’s assessment. As a person ages and accumulates new diagnoses and test results, the model can update its predictions.\n\nThat makes it fundamentally different from the experience most people have with medical risk today. A risk calculation is often performed when a patient is being evaluated for a particular condition. Aladynoulli is designed to keep looking across the record, searching for what might come next.\n\nAnd, in the researchers’ tests, it did that better than several established risk models. Aladynoulli produced more accurate 10-year cardiovascular predictions than three commonly used models, PCE, QRISK3 and PREVENT. It also outperformed the Gail model, a widely used breast cancer risk calculator.\n\nFor breast cancer, Aladynoulli’s 10-year area-under-the-curve score was 0.674, compared with 0.543 for the Gail model. The measurement reflects how well a model distinguishes people who eventually develop a disease from those who do not. It is a measure of predictive discrimination, not a claim that the AI is simply “67.4 percent accurate.”\n\nThe researchers found that Aladynoulli could identify patients at high risk of developing colorectal cancer in the coming year. That could allow a physician to consider additional screening for someone who does not yet meet traditional age-based guidelines.\n\nMedicine has traditionally divided patients into diseases. Aladynoulli tries to reverse that process, starting with the patient and asking how the diseases fit together.\n\n“Two patients with the same diagnosis are not the same patient,” said Pradeep Natarajan, director of Preventive Cardiology at Mass General Brigham. “This model shows they often have different underlying signature profiles, which can translate into different progression patterns and different responses to the same treatment.”", "url": "https://wpnews.pro/news/your-medical-history-may-soon-tell-ai-what-happens-next", "canonical_source": "https://techstrong.ai/articles/your-medical-history-may-soon-tell-ai-what-happens-next/", "published_at": "2026-08-21 15:25:44+00:00", "updated_at": "2026-08-21 15:43:45.719088+00:00", "lang": "en", "topics": ["artificial-intelligence", "machine-learning", "ai-research", "ai-products"], "entities": ["Harvard Medical School", "Dana-Farber Cancer Institute", "Mass General Brigham", "Aladynoulli", "Alexander Gusev", "Giovanni Parmigiani", "UK Biobank", "All of Us Research Program"], "alternates": {"html": "https://wpnews.pro/news/your-medical-history-may-soon-tell-ai-what-happens-next", "markdown": "https://wpnews.pro/news/your-medical-history-may-soon-tell-ai-what-happens-next.md", "text": "https://wpnews.pro/news/your-medical-history-may-soon-tell-ai-what-happens-next.txt", "jsonld": "https://wpnews.pro/news/your-medical-history-may-soon-tell-ai-what-happens-next.jsonld"}}