AI Model Forecasts Risk of More Than 1,000 Diseases Decades in Advance
Key Takeaways:
- Delphi-2M, a generative AI model, can predict susceptibility to over 1,000 diseases using anonymised medical records.
- The system has been tested successfully across large-scale UK and Danish datasets, showing remarkable accuracy and transferability.
- Experts say the model could transform population health forecasting within years and may eventually be adapted for personalised clinical use.
European scientists build AI model to predict long-term disease risk
European researchers have unveiled a powerful new artificial intelligence model that can predict a person’s susceptibility to more than 1,000 diseases decades before symptoms arise.
The system, known as Delphi-2M, was developed by scientists at the European Molecular Biology Laboratory (EMBL) in Cambridge. “Delphi uses a similar architecture to large language models but with key innovations to work with healthcare data,” explained Tom Fitzgerald of EMBL.
Delphi was trained on anonymised health records from 400,000 participants in the UK Biobank, a major long-term biomedical study. The researchers then validated its performance using records from 1.9 million patients in the Danish National Patient Registry.
Matching and exceeding existing prediction tools
The predictions made by Delphi-2M spanned more than 1,000 diseases and were generally comparable in accuracy to existing clinical tools that focus on specific conditions, such as the QRisk score used for cardiovascular disease risk. Results from the study were published in Nature on Wednesday.
“Our model is a proof of concept, showing that it’s possible for AI to learn many of our long-term health patterns and use this information to generate meaningful predictions,” said Ewan Birney, EMBL’s interim executive director. “We were surprised at how well the model transferred from the UK to Denmark though it had never seen a single bit of Danish data.”
Birney emphasised that turning Delphi into a clinically deployable forecasting tool could take five to ten years, but he noted that it could be used much sooner to inform public health strategies.
Population-level insights and healthcare planning
While Delphi generates predictions at the level of individual patients, its most immediate value may be in population health planning. “Although it makes predictions for each individual, it can be very useful at the population level to forecast collective healthcare needs, how many people will suffer from particular diseases such heart attacks, cancers or diabetes and what sort of treatment they need,” said Moritz Gerstung, head of AI at the German Cancer Research Center in Heidelberg and a member of the Delphi team.
The model performed best for diseases with well-understood and consistent progression patterns, such as cardiovascular disease, diabetes and sepsis (blood poisoning). It was less effective for conditions triggered by unpredictable environmental factors or for very rare congenital disorders.
Expanding to genomics and biological data
Researchers are now working to enhance Delphi by incorporating biological information, such as genomic and proteomic data. Despite this, Birney said they were “very pleasantly surprised” at how well the model performed using healthcare records alone, achieving results comparable to or better than some models that rely on genetic and protein-level data.
“I want to stress the power of the straightforward medical record,” Birney added.
The team has patented key aspects of Delphi’s approach to predicting disease risk and timing. “We are exploring whether there are commercialisation possibilities and how to do that with our respective institutions,” Birney confirmed.
Towards ethical and scalable predictive medicine
Independent experts have praised the work as an important step forward for responsible AI in medicine. “This research looks to be a significant step towards scalable, interpretable, and — most importantly — ethically responsible form of predictive modelling in medicine,” said Gustavo Sudre, professor of genomic neuroimaging and AI at King’s College London, who was not involved in the study.
He added that while the current model relies solely on anonymised health records, its architecture has been designed to handle richer data types in the future, including biomarkers, imaging and genomics.
Read More
New AI Tool Pinpoints Genes and Drug Combinations That Restore Health in Diseased Cells
Key Takeaways:
- Harvard researchers have developed PDGrapher, an AI tool that identifies genes and drug combinations most likely to restore diseased cells to a healthy state.
- The model uses graph neural networks to map cellular relationships and predict effective single or combined drug targets, significantly reducing the need for exhaustive drug screening.
- PDGrapher demonstrated high accuracy and speed across multiple cancer datasets, offering promise for personalised medicine and drug discovery in complex diseases such as cancer, Parkinson’s, and Alzheimer’s.
Introduction: A new era in drug discovery
Researchers at Harvard Medical School (HMS) have unveiled a powerful artificial intelligence (AI) model capable of identifying therapies that can reverse disease at the cellular level. This innovation, published in Nature Biomedical Engineering on 9 September, could reshape how new drugs are discovered, designed, and personalised for patients.
Unlike traditional approaches that examine one protein target or drug candidate at a time, this new model — named PDGrapher and freely available to researchers — analyses multiple cellular drivers of disease. It identifies the genes most likely to return diseased cells to normal function and pinpoints the most promising single or combined drug targets to correct the underlying cellular dysfunction.
“Traditional drug discovery resembles tasting hundreds of prepared dishes to find one that happens to taste perfect,” explained senior study author Marinka Zitnik, Associate Professor of Biomedical Informatics at HMS’s Blavatnik Institute. “PDGrapher works like a master chef who understands what they want the dish to be and exactly how to combine ingredients to achieve the desired flavour.”
Limitations of traditional drug discovery
Historically, drug discovery has focused on activating or inhibiting a single protein target. This approach has produced successful therapies such as kinase inhibitors, which block proteins that drive cancer cell growth. However, Zitnik emphasised that such strategies often fall short for diseases driven by multiple interacting pathways and genes.
She noted that many recent therapeutic breakthroughs — including immune checkpoint inhibitors and CAR T-cell therapies — work by targeting broader disease processes rather than single molecules. PDGrapher aims to expand this concept by identifying drug targets that can reverse signs of disease, even when the precise molecular mechanisms are not yet fully understood.
How PDGrapher works: Mapping complex cellular networks
PDGrapher is based on a graph neural network — a form of AI that analyses not only individual data points but also the relationships and interactions between them. In biological research, this means mapping how genes, proteins, and signalling pathways influence one another inside a cell.
Instead of screening thousands of compounds blindly, PDGrapher simulates what would happen if certain genes or pathways were switched off, dialled down, or targeted with a drug. It then predicts whether these interventions would shift a diseased cell towards a healthier state.
“Instead of testing every possible recipe, PDGrapher asks: ‘Which mix of ingredients will turn this bland or overly salty dish into a perfectly balanced meal?’” said Zitnik.
Testing and validation: Proving its predictive power
To train the model, researchers fed PDGrapher a dataset of diseased cells both before and after treatment, allowing it to learn which gene changes led to recovery.
They then evaluated the tool using 19 independent datasets across 11 cancer types, combining both genetic and drug-based experiments. PDGrapher was asked to propose treatment options for samples and cancer types it had never seen before.
The model accurately predicted known drug targets that had been deliberately excluded during training and identified new candidates supported by emerging evidence. Notably, it highlighted KDR (VEGFR2) as a target for non-small cell lung cancer, consistent with clinical findings, and identified TOP2A, an enzyme already targeted by chemotherapy, as a promising target for preventing metastasis in certain tumours.
PDGrapher consistently outperformed comparable AI models — ranking correct therapeutic targets up to 35 percent higher and producing results up to 25 times faster.
Implications for future drug discovery
By focusing on targets that directly reverse disease traits, PDGrapher streamlines the drug discovery process. This allows researchers to prioritise fewer, more promising interventions and to design experiments that are faster and more cost-effective.
This capability is particularly valuable for complex diseases such as cancer, where tumours often evade therapies that strike only one target. Because PDGrapher identifies multiple disease drivers, it offers a way to design combination treatments that could prevent drug resistance.
In the future, with further validation, PDGrapher could be applied to individual patients’ cellular profiles to create personalised treatment strategies.
Broader applications and ongoing research
Beyond cancer, the research team is using PDGrapher to investigate neurological conditions such as Parkinson’s disease and Alzheimer’s disease, aiming to identify genetic drivers that could restore neuronal health.
They are also collaborating with Massachusetts General Hospital’s Center for X-linked Dystonia-Parkinsonism (XDP) to map potential drug targets for this rare, inherited neurodegenerative disorder.
“Our ultimate goal is to create a clear road map of possible ways to reverse disease at the cellular level,” Zitnik stated.
Read More