Using Epigenetic Clocks and EpiScore to Identify Risk and Guide Prevention of Age-Related Diseases

Early epigenetic clocks developed by Horvath [1] and Hannum [2] aimed to predict chronological age with high accuracy using Elastic Net regression. The Horvath clock, based on 353 CpG sites, is characterized by its applicability across multiple tissues. However, these models were not optimized for predicting health outcomes and have limitations in their clinical application.

To overcome these limitations, second-generation clocks such as PhenoAge [3] and GrimAge [4] were developed. PhenoAge incorporates clinical biomarkers like albumin, creatinine, and CRP, while GrimAge includes methylation proxy indicators of plasma proteins and smoking history. These models show a higher correlation with disease onset and mortality risk and have been shown to be effective in stratifying cardiovascular disease, type 2 diabetes, cancer, and dementia [5].

A newer model, DunedinPACE [6], aims to quantify the “pace of aging” rather than biological age. Based on longitudinal cohort studies, it is designed to measure accumulated functional decline.

Fourth-generation approaches are advancing with the integration of tissue-specific methylation profiles and multi-omics information, expected to deepen our understanding of disease specificity and causal relationships.

For example, Polycomb-associated clocks [7] utilize increased DNA methylation at PRC2 (Polycomb Repressive Complex 2) target sites, reflecting tissue-specific aging processes and sensitivity to epigenetic reprogramming. These clocks are mechanistically linked to important biological processes associated with aging, such as tumor formation and the suppression of differentiation.

The difference between epigenetic age and chronological age (Epigenetic Age Acceleration = EAA) has gained attention as a predictor of health outcomes. Large-cohort studies such as the Framingham Heart Study and Generation Scotland have reported associations between EAA and outcomes such as frailty, metabolic syndrome, cardiovascular events, mortality, and cognitive impairment [8,9,10].

Each clock reflects different biological pathways. For example, GrimAge [11] is associated with inflammation and hematopoietic aging, PhenoAge [12] with immune aging and metabolic stress, and DunedinPACE [13] with the accumulation of functional burden. This diversity allows researchers and clinicians to select clocks based on their specific objectives.

While epigenetic clocks are useful, they also face several technical challenges. These include technical variations between platforms such as Illumina 450k and EPIC arrays, batch effects, and confounding due to differences in cell type composition.

Additionally, many models have been trained on data from Western populations, presenting challenges in terms of their generalizability to different ethnic groups. For their future widespread adoption, the development of ethnicity-specific models and standardized analytical methods is essential.

The Horvath Clock (2013) was the first multi-tissue compatible epigenetic clock, aiming to predict chronological age across various human tissues [14]. Its strengths lie in its cross-tissue applicability and high correlation with chronological age, making it useful as a baseline for biological age estimation. The Hannum Clock was also trained to predict chronological age with high accuracy using blood samples [15]. However, both have limited power in predicting functional decline and disease onset, presenting limitations for clinical risk assessment.

PhenoAge was developed by Levine et al. and trained based on clinical biomarkers associated with morbidity and mortality [15]. It excels in evaluating functional aging and is also associated with frailty, inflammation, and immune decline. GrimAge incorporates methylation proxy indicators of plasma proteins (GDF-15, PAI-1, leptin, etc.) and smoking history [16], demonstrating an excellent performance in predicting time to death, cardiovascular events, and cancer onset.

The DunedinPACE clock was developed from longitudinal data based on a birth cohort and reflects the speed of aging rather than accumulated age [17]. It is designed to capture changes that occur over short periods due to lifestyle interventions or treatments and is widely used in recent intervention studies, clinical trials, and behavioral aging research.

Each clock reflects different biological pathways. GrimAge is associated with the blood hematopoietic system and inflammation, PhenoAge with immune aging and metabolic stress, and DunedinPACE with functional decline and multi-disease comorbidity. Which clock to use should be selected according to its purpose, whether for individual health monitoring, population-based risk stratification, or the evaluation of anti-aging interventions (Figure 1).

EpiScores are constructed using machine learning on large-scale epidemiological datasets, aiming to predict protein concentrations, cytokines, and clinical biomarkers from DNA methylation data. Unlike clocks aimed at age estimation, EpiScores focus on predicting disease states, physiological characteristics, and environmental exposure. They have been validated in several cohorts, with high predictive accuracy reported [19].

The integration of EpiScores with transcriptomics, proteomics, and metabolomics enables the elucidation of disease mechanisms and the identification of intervention targets. For instance, combining epigenetic clocks with proteomic markers has been reported to improve prediction accuracy for frailty, cognitive decline, and multi-disease comorbidity [19,21].

EpiScores visualize “hidden risks” before clinical symptoms appear, aiding in the personalization of behavioral changes and preventive measures. In regional public health, they can evaluate the accumulation of biological stress in populations and optimize resource allocation and intervention-planning. They are expected to be integrated with mobile apps and electronic health records (EHR) for use in daily health monitoring and feedback systems.

Multiple studies have shown associations between epigenetic age acceleration and cognitive impairment. Individuals with a high GrimAge or PhenoAge tend to show memory decline, executive function impairment, and an early decline in age-related cognitive function [16,22]. These also correlate with brain atrophy and neuroinflammation, characteristics of Alzheimer’s disease.

Recently, the development of brain-specific methylation clocks has advanced, offering hope for more tissue-specific predictions. Furthermore, integration with polygenic risk scores and neuroimaging (e.g., hippocampal volume and tau PET) may enable the early detection of preclinical Alzheimer’s disease [23].

Cancer risk increases with age, and epigenetic abnormalities are central elements in tumor formation. The acceleration of GrimAge is associated with the onset of lung, breast, colorectal, and prostate cancers [16,24]. These methylation changes reflect the processes of tumor formation, such as chronic inflammation, immune evasion, genomic instability, and cellular senescence.

Additionally, EpiScores specialized in pathways such as p53 signaling, DNA damage response, and angiogenesis enable the more precise stratification of cancer risk. Furthermore, their combination with circulating tumor DNA (ctDNA) and exosome analysis opens new possibilities for minimally invasive cancer diagnosis and treatment monitoring.

Epigenetic clocks show the strongest and most consistent evidence of an association with cardiovascular disease (CVD). GrimAge and PhenoAge strongly correlate with atherosclerosis, coronary artery calcification, left ventricular hypertrophy, and risks of myocardial infarction and heart failure [16,25].

In particular, EpiScores based on biomarkers such as leptin, GDF-15, and PAI-1 enhance the prediction accuracy of cardiovascular events beyond traditional risk models such as the Framingham score. This greatly contributes to preventive medicine by enabling early intervention at asymptomatic stages.

In countries like Japan that have advanced health screening systems yet still have cardiovascular disease as a major cause of death, incorporating biological aging markers into health examinations would strengthen primary prevention and risk stratification.

Despite their growing utility, epigenetic clocks and EpiScores face several limitations. These include inter-platform variability (e.g., 450K vs. EPIC 850K arrays), poor reproducibility across laboratories, and tissue-specific differences in accuracy. Additionally, clocks trained in predominantly European populations may underperform in non-European groups. EpiScores offer disease-specific insights but are not yet widely validated or cost-effective. Integration into clinical systems remains limited due to a lack of standardization and regulatory oversight. Comparisons with protein biomarkers and polygenic risk scores (PRS) show mixed results, suggesting the need for multi-modal approaches. Broader efforts in international standardization, such as ISO-level guidelines and ethnicity-specific model development, are ongoing and essential for widespread adoption [26].

In Japan, specific health examinations are mandatory for those aged 40–74, aiming to achieve the early detection of metabolic syndrome and to prevent lifestyle-related diseases. However, this screening system mainly relies on standard indicators, such as BMI, blood pressure, and cholesterol, making it difficult to capture early biological changes associated with aging.

Incorporating “biological age” into such systems greatly improves the accuracy of risk assessment. For example, a 50-year-old with a biological age of 60 may be judged to need stronger intervention than someone with a biological age of 45. Feedback like “your biological age is 10 years older than your actual age” is intuitively easy to understand and serves as powerful motivation, especially for asymptomatic individuals.

Healthcare professionals such as public health nurses, primary care physicians, and dietitians can use this information to personalize lifestyle improvement guidance. In fact, counseling using biological age has been reported to improve patient engagement and intervention continuation rates.

One of the most promising applications of epigenetic clocks is evaluating intervention effects. Multiple studies have shown that interventions in diet, exercise, sleep, and stress management can delay or even rejuvenate biological age [25,27,28]. For instance, caloric restriction and aerobic exercise have been confirmed to lower epigenetic age within months, as measured by GrimAge and DunedinPACE.

By measuring biological age before and after interventions, healthcare providers and patients can objectively confirm the effects of interventions. This promotes behavioral change and enables the redesign of personalized strategies. The longitudinal tracking of aging indicators also serves as a guide to determine whether intensification or adjustment of the intervention is necessary [29].

With the decreasing costs of methylation testing and increasing test availability, the feasibility of implementing personalized preventive medicine continues to expand. Integration with electronic health records (EHR), wearable devices, and mobile health apps enables the continuous monitoring of biological age and real-time feedback.

Ultimately, combining epigenetic clocks with EpiScores and multi-omics information (microbiome, transcriptome, metabolome, etc.) will lead to more comprehensive individual profiling. Such efforts support the transition from conventional “medicine that treats after symptoms appear” to a “proactive approach to maintain biological function”. This is an essential shift, especially for super-aged societies like Japan.

Longevity clinics in America, Japan, Singapore, UAE, and other countries have introduced biological age assessments using epigenetic clocks as part of comprehensive health examinations. Patients undergo DNA methylation testing and receive individualized recommendations on diet, exercise, supplements, and regenerative medicine based on the results.

Testing companies with CLIA certification, such as TruDiagnostic and Zymo Research, contribute to ensuring clinical reliability and international comparability. In Japan, some clinics in Tokyo and Osaka have begun incorporating epigenetic clocks into standard anti-aging programs, conducting longitudinal follow-ups and measuring the effects of lifestyle interventions.

Such applications are not just privileges for the wealthy. With declining testing costs and increasing social interest, their use is spreading to wellness-oriented citizens and the working generation in their 40s to 60s.

In aging societies like Japan, the significance of biological age measurement is increasing in the context of comprehensive community care and frailty prevention. For example, even elderly people who appear healthy may benefit from early intervention (resistance training, nutritional support, social participation programs, etc.) if their biological age is significantly advanced.

Additionally, biological age feedback enhances the effectiveness of the behavioral change support provided by public health nurses, dietitians, and municipal staff. Biological age is more intuitively understandable than abstract biomarkers and serves as an easily usable indicator for patients’ behavioral guidance.

Some municipalities have begun pilot projects that link biological age measurement, though not at the epigenetic clock level, with regional wellness platforms and incentive systems. For example, models in which residents can earn rewards by engaging in health behaviors linked to health point systems are being tested.

The use of epigenetic clocks is also being considered as an objective intervention criterion for frailty prevention in nursing care. This would make it possible to allocate care prevention resources more accurately.

A notable case is the Osaka-Kansai Expo 2025. The Osaka Pavilion plans to offer visitors the opportunity to visualize their biological age on the spot through non-invasive tests like grip strength, body composition balance, dental alignment, and cognitive measurements [30]. This represents an important step in spreading a new perspective in society: not “How old are you?” but “How healthily are you aging?”

Figure 2 provides a synthesized framework illustrating how EpiScores and epigenetic clocks can be integrated into clinical practice and public health systems, aligning with the discussed implementation strategies.

As electronic health record (EHR) systems expand their functionality to handle genomic, epigenomic, and multi-omics data, the introduction of epigenetic age is reaching a realistic stage. Biological age can be incorporated into daily clinical practice as a new “next-generation vital sign”, in addition to conventional vital signs such as blood pressure, cholesterol, and BMI.

For example, patients with advanced biological age despite normal test values are identified as candidates for early intervention against hidden risks. Conversely, for individuals showing healthy aging, examination frequency and intervention levels can be flexibly adjusted.

Additionally, by adding multi-omics indicators (e.g., gut microbiota, blood metabolites, and inflammation markers), more precise health profiling and predictive models are being developed within EHRs.

Mobile health (mHealth) technology and wearable devices provide a new means to visualize the interface between biological age and daily life in real-time. Individuals can import their DNA methylation test results into apps and receive dynamic reports and behavioral suggestions.

By providing feedback on the relationship between daily life factors such as physical activity, sleep, stress, nutrition, and biological age or EpiScores, users can “visualize” the impact of their lifestyle habits on aging. This serves as an extremely powerful motivation for behavioral change.

Furthermore, more comprehensive and self-directed health coaching becomes possible by combining self-assessments based on the Intrinsic Capacity framework (e.g., mood, vitality, and concentration).

Several challenges exist in the social implementation of such systems. These include the standardization of measurement protocols, interoperability between platforms, legal and ethical regulations, and the establishment of privacy protection. An intuitive and interpretable feedback design for both healthcare providers and patients is also necessary.

Nevertheless, as precision medicine and public health converge, biological age indicators will become central to dynamic and predictive health management through the coordination of multi-omics and AI technology.

In the future, such indicators will function as a “comprehensive aging dashboard” integrated with Intrinsic Capacity and subjective well-being, making it realistic for individuals and society to co-design a healthy life expectancy.

AI possesses the ability to process high-dimensional, multi-layered data such as DNA methylation, transcriptomics, proteomics, metabolomics, and life logs. Machine learning algorithms elucidate complex relationships between specific epigenetic signatures and disease risk, aging rate, and health outcomes.

For example, deep learning models can automatically estimate biological age from raw methylation β values, enabling risk stratification in populations. Based on longitudinal data, predictions of aging trajectories and intervention simulations are also possible.

Furthermore, AI supports the construction of individual aging profiles and digital twins (in silico reflection of lifestyle and treatment effects). This enables the realization of biologically personalized and dynamically adaptive prevention strategies.

The exposome refers to the cumulative environmental exposures (air pollution, diet, psychological stress, chemicals, infections, and lifestyle habits) throughout life. These interact with the genome and epigenome, shaping health trajectories over decades.

Epigenetic clocks show sensitivity to many elements of the exposome, including PM2.5, noise, nutritional deficiencies, childhood adversity, and occupational exposures. Specific DNA methylation sites can serve as biomarkers for exposure to toxins and chronic stress.

EpiScores reflecting inflammation, oxidative stress, and metabolic abnormalities can also function as intermediate phenotypes between the exposome and disease. The integration of exposome and epigenetics enables an understanding of the environmental origins of aging and identification of modifiable risk factors.

Combining AI and exposome analysis with epigenetic clocks enables a transition from “one-size-fits-all” prevention to stratified population prevention. This approach allows for the early identification of high-risk individuals and the more equitable distribution of public resources.

In the future, digital platforms are expected to provide users with personalized advice based on cumulative exposure history and biological age. At the municipal and national levels, these tools can be used to understand population aging trends and health disparities, optimizing policy development.

International organizations and foundations such as the World Health Organization (WHO), Hevolution Foundation, and XPRIZE for Healthspan are working to establish advanced frameworks that view aging as a “measurable and modifiable biological process”. While global consensus has not yet been reached, these movements are noteworthy for their focus on visualizing and preventing aging.

These efforts aim to validate epigenetic clocks across ethnic and regional differences, conduct long-term longitudinal tracking of aging interventions, and build international consortia for data-sharing.

PRC2-AgeIndex, GrimAge, and DunedinPACE have shown promising results in international validation studies, but challenges remain in terms of region-specific adjustments. The importance of local adaptation, particularly in Asia, Africa, and Latin America, is crucial to avoid widening health disparities.

Incorporating biological age into healthcare systems brings several ethical and legal challenges, such as the following: Should insurance premiums vary according to biological age? How should biological age be utilized in workplace health assessments? What protective measures are needed to prevent age-based discrimination?

In response to these concerns, the international bioethics community is developing frameworks that emphasize privacy, fairness, and empowerment. Biological age information should be used to support individual care and autonomy, not to disadvantage individuals based on factors beyond their control.

Additionally, the regulatory approval of diagnostic tools based on methylation testing (such as CLIA in the US, CE-IVD in Europe, and PMDA in Japan) is also evolving to reflect the clinical significance of aging biomarkers.

With the progression of global aging, new frameworks are needed to scientifically and equitably extend healthy life expectancy. The construction of a “Global Health and Aging Index” or “Well-being Score” that integrates biological age measurement technologies (epigenetic clocks, EpiScores), functional indicators, environmental factors, and subjective well-being assessments is attracting attention as the next stage.

Such indicators are very useful for various purposes, including the following: Real-time understanding of the aging status of populations in each country and region. Early stratification of disease onset risk and prioritization of appropriate interventions. Scientific and equitable allocation of healthcare, nursing care, and health resources. “Visualization” of health disparities due to social and economic factors.

For example, Intrinsic Capacity, proposed by WHO, is a concept that comprehensively captures the physical, mental, and social functions of the elderly and is gaining international traction. It is also consistent with the European Union’s “Healthy Aging” policy and aging response measures in Asian countries, gradually paving the way for the international standardization of comprehensive aging indicators.

In Japan, the “Visualization of Aging” project at the Osaka-Kansai Expo 2025 is attracting attention. Visitors will gain experience confronting their biological age through epigenetic clocks and well-being scores [31]. This serves as an opportunity to transform the perception of aging from “invisible” to “visible,” and then to “changeable.”

The ultimate goal is to balance evidence-based health policies and personalized preventive interventions. Building a measurable, modifiable, and equitably distributable paradigm of healthy longevity through collaboration between government, academic institutions, industry, and civil society will be an important mission in 21st-century public health.