AI Model Uses De-Identified Patient Data to Predict Healthcare Needs in Pilot Study

A new pilot study has demonstrated the potential of an advanced AI system named Foresight, designed to analyze de-identified patient data and forecast future healthcare events. This model functions similarly to popular language models like ChatGPT by learning to predict upcoming medical outcomes based on previous health records. The model is trained exclusively on routinely collected NHS data, such as hospital admissions and vaccination rates, covering the entire population of England.
Foresight aims to identify individuals at high risk of events like hospitalization, heart attacks, or new diagnoses, enabling healthcare providers to intervene early. Conducted within the secure NHS England Secure Data Environment (SDE), the study ensures strict privacy controls by maintaining all patient data under NHS oversight. This approach allows predictions to incorporate diverse demographic groups across England, including rare conditions and minority populations.
Led by Dr. Chris Tomlinson from UCL, the research emphasizes the importance of representing the full spectrum of the population in training data to improve model accuracy and inclusivity. The British Heart Foundation Data Science Center facilitated access to the SDE, with public involvement in the approval process to uphold transparency.
Experts believe that Foresight's predictive capabilities could significantly enhance preventative healthcare strategies and help address healthcare inequalities. The model’s potential to analyze population-wide health risks supports NHS planning and resource allocation.
Researchers like Simon Ellershaw highlight the technical achievements made possible by collaboration with NHS partners, enabling secure application of cutting-edge AI at an unprecedented scale. The current pilot uses recent data (2018-2023) focused on COVID-19 research, with plans to expand to deeper and richer data sources in future iterations.
Prominent figures, including Professor Richard Dobson of King’s College London, underline the importance of incorporating more detailed data such as clinical notes and test results. They stress the necessity of keeping patient and public interests at the forefront, ensuring ethical use of AI in healthcare.
The UK government is supporting this initiative through the development of a Health Data Research Service, promoting secure data access for health innovation. Officials like NHS Transformation Director Dr. Vin Diwakar and officials from the Department of Health emphasize that AI-driven predictive models could transform disease prevention and treatment, enhance personalized care, and accelerate UK’s leadership in trustworthy AI applications in health.
This pioneering research showcases the UK’s commitment to leveraging AI for better health outcomes while maintaining high standards of data privacy and ethical oversight.
source: https://medicalxpress.com/news/2025-05-ai-de-patient-health.html
Stay Updated with Mia's Feed
Get the latest health & wellness insights delivered straight to your inbox.
Related Articles
Extended Overweight or Obesity During Pediatric Leukemia Treatment Predicts Poorer Outcomes
New research shows that prolonged overweight or obesity during pediatric leukemia treatment is linked to higher relapse risk and lower survival. Managing weight during therapy can improve outcomes.
TikTok Skin-Care Routines in Teens May Increase Risk of Chronic Allergies
A recent study warns that TikTok skin-care routines among teens may cause lifelong allergies and skin conditions due to high-risk ingredients and unregulated practices. Learn more about the potential health implications.
Research Finds Early Preemptive Kidney Transplantation Does Not Lower Mortality Risk
A Yale University study finds that early preemptive kidney transplants do not lower mortality risk, suggesting waiting until transplant is needed may be more beneficial.