Dynamic Disease Prediction

In age-stratified and longitudinal cohorts, DNA methylation patterns provide insight into how disease trajectories and mortality risk evolve over time. Methylation Risk Scores (MRS) can be applied across multiple chronic conditions, supporting more accurate patient stratification and earlier identification of clinically meaningful risk differences. Analyses using identical Generation Scotland datasets from Harvard demonstrate that:

‍

• MRS more effectively differentiate patient outcomes in conditions including Type-2 Diabetes, Atrial Fibrillation, and Myocardial Infarction
• Methylation-based models improve screening performance by reducing false positives
• Compared with established risk models published in leading journals over the past 5 years, these approaches demonstrate superior performance in approximately 73% of comparisons

‍

For clinicians and researchers focused on confident risk stratification, these models enable highly specific patient categorization, with positive predictive values exceeding 90% in selected endpoints. This level of precision supports more informed long-term planning for chronic disease management.

‍

Carreras Gallo N, et al. "Leveraging DNA methylation to create Epigenetic Biomarker Proxies that inform clinical care: A new framework for Precision Medicine." medRxiv preprint. 2024. doi:10.1101/2024.12.06.24318612

DNA methylation (DNAm) patterns can reveal molecular signatures associated with disease processes before they are detectable through conventional clinical observation. For example, DNAm-based models have demonstrated utility in conditions such as schizophrenia, where diagnosis has historically relied on symptomatic presentation rather than biological measures.

‍

In other diseases, such as Chronic Obstructive Pulmonary Disease (COPD), validated predictive diagnostics remain limited, constraining opportunities for proactive intervention. Methylation-based models offer a path toward earlier biological risk identification and improved preventive strategies.

‍

Because DNAm reflects ongoing biological response to environmental exposures, behavior, and nutrition, these models can update dynamically, enabling clinicians to monitor changes in risk and adjust care pathways accordingly.

Methylation Diagnostic Scores (MDS) are also well suited for prognosis forecasting. By enabling tighter stratification of disease trajectories, MDS provide clinicians with a more precise, forward-looking view of how individual patients are likely to progress under different treatment scenarios. This is particularly valuable in areas such as oncology, where understanding whether treatment is likely to be required within two years versus ten can meaningfully alter clinical decision-making. MDS-based stratification supports more informed planning around treatment timing, intensity, and sequencing. For prognosis, MDS can:

‍

• Improve forecasting of disease trajectory over extended time horizons
• Reduce unnecessary treatment exposure and associated adverse effects
• Support real-time monitoring and data-informed course correction as therapy continues

‍

More accurate prognosis ultimately contributes to better patient outcomes and more efficient use of healthcare resources. Methylation-based diagnostic matching offers a framework for improving treatment fit, reducing unnecessary trial-and-error, and shortening the time to optimal therapy.

‍

Getting asthma therapy right


In one study, mismatched asthma treatments were associated with an additional $10,000–$50,000 in annual cost per patient. Precision matching using methylation-based diagnostic models reduced unnecessary spend, saving up to $50,000 per patient per year, while also lowering hospitalization rates.


‍

Example provided to illustrate economic impact; results vary by indication.

As methylation datasets expand, they enable increasingly refined characterization of patient typologies and response patterns. Larger and more diverse datasets allow researchers to disentangle complex, interacting biological signals and better identify which interventions drive meaningful outcomes.
Growing datasets enable:

‍

• Richer comparative analyses across patient subgroups
• More granular prediction of likely outcomes
• Improved identification of high-impact therapeutic and behavioral interventions


‍

Recent studies leveraging longitudinal methylation data have successfully quantified the biological effects of interventions and health behaviors over time using MRS and MDS. As these datasets continue to scale, they offer a powerful foundation for linking epigenomic patterns with disease risk, preventive strategies, and long-term health trajectories.

Methylation Risk Scores enable companion diagnostic approaches that move beyond generalized risk estimates, supporting drug-specific prediction of treatment response and more efficient allocation of therapies.

‍

‍GLP-1 response prediction

‍Challenge: Response to GLP-1 receptor agonists varies widely across patients. Some individuals experience limited efficacy, while others may develop adverse effects such as disproportionate lean muscle mass loss.


‍
‍Approach: Methylation Risk Scores (MRS) identify patient typologies associated with differential treatment response using drug-specific epigenomic models.


‍Key findings: ‍

• Drug-specific methylation models outperform generic risk stratification approaches
• Enable pre-treatment identification of likely responders and non-responders
• Provide greater individual-level response prediction than aging clocks alone

‍

Economic implications: Precision diagnostic matching using MRS can reduce healthcare costs by:

• Eliminating ineffective treatment cycles
• Reducing management of avoidable adverse events
• Shortening time to optimal therapy
• Lowering hospitalization risk

  ‍

MRS and biological aging clocks function as complementary tools: aging clocks provide a broad assessment of overall health status, while MRS support treatment- and disease-specific prediction.

‍

Corley MJ, et al. "Semaglutide Slows Epigenetic Aging in People with HIV-associated Lipohypertrophy: Evidence from a Randomized Controlled Trial." medRxiv preprint. 2025. doi:10.1101/2025.07.09.25331038