A new AI model called Delphi-2M has been developed to predict the likelihood of over 1,000 diseases developing within the next two decades by analyzing medical histories, lifestyle factors, and health conditions.

While it performs well for diseases with predictable progression like cancer and heart attacks, its reliability drops for more variable conditions such as mental health disorders and pregnancy complications.

The model was trained on anonymized data from 400,000 UK Biobank participants and 1.9 million Danish patients, achieving high accuracy comparable to existing single-disease models.

Experts caution that Delphi-2M is not yet ready for clinical use due to biases in the training data related to age, ethnicity, and health outcomes, and it requires further testing and validation.

The model's development adhered to strict ethical standards, including secure data handling and informed consent, with privacy protections in place by keeping data within national borders.

Before clinical deployment, the model needs further testing, regulation, and refinement, with experts comparing its potential to the decade-long integration of genomics into healthcare.

This innovation signals a shift toward predictive and preventive healthcare, but challenges such as explainability, privacy, and clinical integration still need to be addressed.

Future enhancements may include integrating molecular and wearable data to improve the accuracy of long-term health forecasts and expand applicability.

The goal is for Delphi-2M to assist in healthcare planning and resource allocation by providing population-level risk assessments, aiding early intervention strategies.

While promising for early diagnosis and prevention, experts highlight the importance of assessing the psychological impact on patients who learn about their high disease risks.

Current limitations include underrepresentation of childhood health events and certain demographic groups in the training data, which may bias predictions and highlight the need for more diverse datasets.

The model identifies patterns of disease progression from current health data, aiming to help healthcare professionals with early diagnosis and estimating disease burden at a population level.