Concept of an AI Aging Clock Based on ACCO Watches and Accofrisk AI Health Intelligence.

Introduction.

Modern medicine already uses a broad range of aging clocks — scientific models that estimate a person’s biological age, or the biological age of specific organs and physiological systems, using biological, physiological, molecular, or imaging data.

There are already epigenetic clocks based on DNA methylation, as well as proteomic, metabolic, physiological, and imaging-based models. Another important area is the assessment of the biological age of individual systems, including the brain, heart, vascular system, lungs, liver, kidneys, metabolic system, musculoskeletal system, and sleep. Multi-organ biological-age models are also becoming an active area of AI-based aging research.

For ACCO, the most relevant approaches are those for which the required input parameters are already available from ACCO data or can be supplemented with information entered by the user.

It is important to distinguish between two approaches: using established age-estimation methodologies and developing new AI Aging Clock models based on data processed by Accofrisk AI Health Intelligence.

What We Mean by “AI Aging Clock”.

In this article, AI Aging Clock refers to a class of artificial intelligence models that use a combination of physiological and biological data to estimate a person’s biological or functional age, either overall or for specific body systems.

Unlike specific scientific methodologies such as the Epigenetic Clock, Proteomic Clock, or Imaging Aging Clock, the term AI Aging Clock describes an approach rather than a single established methodology: a machine-learning model analyzes multiple parameters, their relationships, and changes over time to generate an age estimate.

In the context of ACCO, we use the term to describe a concept in which data from ACCO watches and Accofrisk AI Health Intelligence could support existing age-estimation methodologies or the development and subsequent scientific validation of proprietary AI Aging Clock models for different body systems.

Age-estimation methodologies that can be implemented using ACCO data.

Framingham Heart Age is an established approach for estimating cardiovascular age and cardiovascular risk. It uses age, sex, total cholesterol, HDL cholesterol, systolic blood pressure (SBP), smoking status, diabetes, and information about antihypertensive treatment. Some of these parameters can be obtained from ACCO, while additional information such as smoking status and medication use can be entered by the user. This makes the methodology potentially implementable using ACCO data, provided that the complete required input set is available.

Fitness Age / Cardiorespiratory Fitness Age refers to established approaches for estimating the age of a person's physical fitness. A key parameter is maximal oxygen uptake (VO₂max), while different models may also use age, sex, body mass index (BMI), body-fat percentage, physical activity, and other measures of physical fitness. ACCO includes oxygen consumption (VO₂), BMI, body-fat percentage, muscle mass, physical activity, workload, and recovery parameters. This makes Fitness Age one of the most natural candidates for implementation using ACCO data.

AI-VascularAge is an AI-based approach to estimating vascular age from pulse-wave characteristics. This is particularly relevant to ACCO because the device captures the pulse wave at the wrist. This creates a potential basis for adapting such an approach to ACCO data. However, the signal quality, measurement technique, data format, and preprocessing would need to be validated against the data used to develop the original model.

Concept of AI Aging Clock models based on data from Accofrisk AI Health Intelligence.

The next level is to use the multidimensional physiological data processed by Accofrisk AI Health Intelligence to develop proprietary models for estimating biological age.

This is not simply a matter of applying someone else's published formula. The concept is to combine established scientific knowledge about age-related changes in organs and physiological systems with data from Accofrisk AI Health Intelligence and machine-learning methods, and then train and validate new models using appropriate clinical and age-stratified datasets.

Brain Age — biological age of the brain. ACCO includes a broad range of parameters related to cerebral hemodynamics and oxygen metabolism, including cerebral blood flow (CBF), cerebral blood volume (CBV), cerebral perfusion pressure (CPP), cerebrovascular reactivity (CVR), intracranial pressure (ICP), cerebral metabolic rate of oxygen (CMRO₂), jugular venous oxygen saturation (SjvO₂), and other parameters. These data could provide a basis for developing a proprietary Brain Aging Clock. However, a published model cannot automatically be transferred if, for example, its CBF measurements were obtained using a specific technique such as ASL-MRI.

Cardiac Age — biological age of the heart. ACCO includes cardiac output (CO), cardiac index (CI), global ejection fraction (GEF), cardiac power output (CPO), stroke work (SW), myocardial oxygen consumption (MVO₂), and other indicators of cardiac function. Their combined use could provide the basis for a proprietary Cardiac Age model.

Vascular Age — biological age of the vascular system. In addition to pulse-wave data, ACCO includes blood pressure, blood-flow, perfusion, microcirculation, blood-viscosity, fibrinogen, and hematocrit parameters. Together, these measurements could support the development of a dedicated Vascular Aging Clock.

Kidney Age — biological age of the kidneys. ACCO includes renal blood flow (RBF), renal filtration fraction (RFF), renal perfusion pressure (RPP), serum creatinine, blood urea nitrogen (BUN), and uric acid. These parameters could be used to develop a model of functional kidney age.

Liver Age — biological age of the liver. ACCO includes alanine aminotransferase (ALT), aspartate aminotransferase (AST), the AST/ALT ratio, ammonia (NH₃), lactate, and metabolic parameters. These measurements could form the basis of a proprietary Liver Aging Clock.

Metabolic Age — metabolic age. This is one of the areas with the richest set of potentially relevant ACCO parameters. These include glucose, glycated hemoglobin (HbA1c), HOMA-IR, HOMA-B, the triglyceride-glucose index (TyG), visceral adiposity index (VAI), lipid accumulation product (LAP), uric acid-to-HDL ratio (UHR), BMI, body-fat percentage, muscle mass, resting metabolic rate (RMR), and other indicators. Metabolic Age could therefore become one of the first proprietary AI Aging Clocks developed using data from Accofrisk AI Health Intelligence.

Lung Age — functional age of the respiratory system. ACCO includes oxygen consumption (VO₂), minute ventilation (VE), alveolar ventilation (VA), respiratory quotient (RQ), arterial oxygen partial pressure (PaO₂), arterial carbon dioxide partial pressure (PaCO₂), physiological dead space (VD), rapid shallow breathing index (RSBI), pulmonary vascular resistance (PVR), mean pulmonary artery pressure (MPAP), and other respiratory and gas-exchange parameters. These data could support a proprietary model of respiratory-system age. However, it should not automatically be called the classical Spirometric Lung Age if the original methodology requires forced expiratory volume in one second (FEV₁).

Musculoskeletal / Functional Age — functional age. ACCO includes muscle mass, body-fat percentage, BMI, resting metabolic rate (RMR), physical activity, workload, and recovery. Together, these parameters could support a model of physical and functional age.

Sleep Age — estimated age of sleep characteristics. ACCO includes sleep duration, sleep stages, apnea, snoring, and other sleep-related characteristics. These data could potentially be used to develop a proprietary Sleep Age model that estimates how closely an individual's sleep profile corresponds to patterns observed at different chronological ages.

What cannot simply be transferred from existing models.

Some aging clocks require completely different types of data. Epigenetic clocks rely on DNA methylation patterns. Proteomic clocks require measurements of specific proteins. MRI-based aging clocks use medical images and quantitative characteristics of tissues and organs.

Therefore, having a physiologically similar parameter does not mean that a published formula can simply be applied to ACCO data. For example, if a Brain Age model was trained using cerebral blood flow (CBF) measured by ASL-MRI, the presence of CBF in ACCO data does not automatically make the original formula applicable. In such a case, the appropriate approach would be to develop and validate a new model using data from Accofrisk AI Health Intelligence.

Overall AI Aging Clock concept.

ACCO could therefore become more than a device for measuring individual physiological parameters. It could serve as a source of multidimensional physiological data for a system of biological-age assessment.

The first level would be the integration of established methodologies:

Framingham Heart Age → Fitness Age → AI-VascularAge

The second level would be the development of proprietary models based on data from Accofrisk AI Health Intelligence:

Brain Age → Cardiac Age → Vascular Age → Kidney Age → Liver Age → Metabolic Age → Lung Age → Musculoskeletal / Functional Age → Sleep Age

These individual indicators could then potentially be integrated into an Overall Biological Age.

Instead of providing only one generalized number — for example, “biological age: 52” — the system could generate an individual map of physiological aging:

This approach would show not only how old the body appears to be overall, but also which physiological systems may be functionally older or younger than the person's chronological age.

That is the core concept of an AI Aging Clock based on ACCO watches and Accofrisk AI Health Intelligence: ACCO watches capture the pulse wave and other physiological parameters; Accofrisk AI Health Intelligence processes these data, analyzes them longitudinally, and uses them to generate individualized indicators of physiological status and functional age across different body systems.

Established scientific methodologies can be used where the ACCO data are sufficiently compatible, while new models can be developed and validated for areas where direct transfer of existing methodologies is not scientifically justified.