How Biological Age Science Has Evolved, and What the Latest Models Can Now Tell You

Introduction: Aging as a Measurable State

Two people can share the same birthday and have almost nothing else in common biologically. One may carry the metabolic profile of someone a decade younger. The other may show early signs of organ stress, systemic inflammation, and declining physiological reserve that place their true health trajectory well ahead of their years. Chronological age, the number on a birth certificate, captures neither of these realities. Biological age attempts to.

The idea driving this field is straightforward, even if its execution is not. If aging leaves measurable traces in the body, those traces can be detected, quantified, and translated into something more useful than a birthday. Blood-based models have emerged as one of the most practical tools for doing this. Routine laboratory markers, the kind collected during a standard clinical panel, encode information about metabolism, organ function, immune activity, and systemic stress that, when analyzed together, can reveal how an individual's physiology compares to population-level aging trajectories.

The field's foundation was laid by the Levine PhenoAge model, a second-generation biological age estimator that demonstrated something important: a small set of standard blood biomarkers could predict mortality risk with meaningful accuracy, capturing biological information that chronological age alone could not [1]. What PhenoAge established in principle, subsequent models have worked to refine in practice.

The most recent of these refinements is the Bortz model, published in 2023 [2]. Developed using larger datasets and more sophisticated statistical methods than its predecessors, it applies a multivariate algorithm to routine blood biomarkers to generate a single, interpretable estimate of biological age. The output is not a prediction of how long someone will live. It is a measure of where their physiology currently sits relative to population-level mortality risk curves, translating complex biomarker patterns into a signal that is both clinically meaningful and practically actionable for anyone serious about monitoring the trajectory of their own aging.

From Estimation to Application and Optimization

Knowing your biological age is only useful if it changes what you do next.

The statistical elegance of a biological age model matters far less than what happens when it meets the real world. A score that sits in a research paper, however accurate, does not extend anyone's healthspan. What does is the ability to translate that score into a clinical signal, identify what is driving it, and act on that information with enough precision to move the needle over time.

This is the gap that Dr. Jim Lanzilotti and Zsolt Szabo, working within the Healthspan framework, are attempting to close. Szabo's work has focused on making the Bortz model practically accessible, translating a sophisticated multivariate algorithm into a format that individuals can apply to their own routine laboratory data. Dr. Lanzilotti's clinical approach takes that translation a step further, embedding biological age monitoring within a broader partnership model that combines strategic laboratory panels, early identification of physiological drift, and targeted intervention plans designed to support long-term healthspan.

What unites these approaches is a shared premise worth stating clearly. No single biomarker, and no single model, captures the full complexity of how a body ages. Biological age is a powerful composite signal, but it is also an incomplete one when read in isolation. Its value emerges when it is interpreted alongside the domain-specific markers that reveal what is driving it, and when that interpretation is connected to a clinical strategy that can act on what the data shows. Together, measurement and interpretation form something closer to a complete picture of where an individual's physiology is heading and how quickly.

The opportunity this creates is not simply to measure aging more precisely, though precision matters. It is to operationalize that measurement, connecting validated research models like Bortz with clinical partnership, lifestyle modification, coaching, and where appropriate, pharmacological support, so that biological age becomes not a verdict but a variable. Something that can be tracked, understood, and actively shaped over time.

A Continuum of Progress: From Levine to Bortz and Beyond

Scientific fields rarely advance in clean leaps. They move through a series of refinements, each generation of tools building on the limitations of the last, each improvement in precision opening new questions that the previous framework could not ask. The evolution of biological age estimation is no exception.

The Levine PhenoAge model marked the critical inflection point. Before it, the idea that routine blood biomarkers could encode meaningful information about mortality risk and translate that risk into a single biological age estimate was largely theoretical. PhenoAge made it practical, providing one of the first validated frameworks for quantifying aging in a clinical context and demonstrating that the signal was real enough to be useful [1].

What followed was an expansion of the paradigm in multiple directions simultaneously. Epigenetic clocks emerged as a parallel approach, reading aging signals directly from patterns of chemical modification to DNA. Proteomic and metabolomic models added further layers, capturing information about cellular regulation, immune function, and metabolic state that blood chemistry alone cannot fully encode [3]. Each of these approaches offered something the others could not, and each came with its own tradeoffs. Greater biological resolution, in most cases, meant reduced accessibility. The models that captured aging most completely were rarely the ones that could be run on a standard clinical panel.

This tension between depth and practicality remains one of the central design problems in the field. A biological age model that requires specialized assays and expensive laboratory infrastructure may be scientifically impressive while remaining clinically irrelevant for the overwhelming majority of people who might benefit from it. The most useful model is not necessarily the most sophisticated one. It is the one that delivers actionable insight from data that is actually obtainable.

The Bortz model sits within this context as a meaningful refinement of earlier blood-based approaches [2]. Developed on larger datasets and with more sophisticated statistical methods than PhenoAge, it improves predictive performance without abandoning the clinically accessible biomarkers that make real-world application possible. It is a better instrument built from the same materials, calibrated with more precision.

Looking further forward, the field is moving toward frameworks that are not only more accurate but more dynamic. Future biological age models may incorporate longitudinal data and adaptive baselines, tracking how an individual's aging trajectory shifts in response to interventions, lifestyle changes, and the passage of time rather than measuring a single static snapshot [5]. The integration of multiple physiological domains into unified frameworks reflects a growing recognition that aging does not unfold along a single pathway but emerges from the interaction of metabolic, immune, hormonal, and cellular systems operating simultaneously [4].

The Bortz model is best understood not as a destination but as a waypoint in this longer journey. It builds on what Levine established, improves on it in ways that matter for clinical use, and points toward a future in which biological age is not a fixed number assigned at a single measurement but a continuously updated signal that responds to how a person lives, what interventions they pursue, and how their physiology evolves over time.

From Model to Measurement: Why Lab Data Matters

A biological age score is only as useful as the information feeding into it. The number itself is the output. What matters for intervention is understanding the physiology generating it.

Think of routine laboratory data as a control panel for aging. Each marker is a readout from a specific physiological system, metabolism, organ function, immune activity, cellular stress, and together they form a systems-level picture of how the body is holding up. The critical feature of this picture is its timing. These markers capture early changes in physiology that often emerge years, sometimes decades, before clinical disease becomes diagnosable. By the time a condition reaches the threshold for a formal diagnosis, the biological drift that produced it has usually been underway for a long time. Lab-based monitoring creates the opportunity to detect and act on that drift earlier, when intervention is most likely to make a difference.

More advanced approaches exist. Epigenetic clocks read aging signals directly from DNA methylation patterns. Proteomic and metabolomic platforms capture biological information at a resolution that routine blood chemistry cannot approach. These tools offer genuine depth, but they remain constrained by cost, specialized infrastructure, and limited clinical scalability. They are valuable research instruments. They are not yet tools that most people can use consistently to monitor and guide their own health trajectory.

Routine laboratory markers occupy a different position. They are accessible, repeatable, and interpretable within a standard clinical relationship. This is the principle that unites the Bortz model, Szabo's translation work, and the Healthspan clinical framework: the best model is not the most sophisticated one. It is the one that can be used consistently, interpreted clearly, and acted upon reliably. By keeping biological age estimation grounded in clinically available markers, this approach brings longevity science out of specialized research environments and into everyday clinical practice, where it can inform real decisions and produce measurable improvements in healthspan for people who would never have access to an epigenetic clock or a proteomics platform.

Their limitation is not sensitivity but context. A single biomarker in isolation tells an incomplete story. Biological age is one layer of interpretation built on top of a broader laboratory foundation that must include focused assessments of liver function, glucose regulation, hormone balance, thyroid signaling, and iron status, among others. Each of these domains adds a dimension of physiological context that the biological age score alone cannot provide, and that is necessary for translating a number into a coherent picture of what is actually happening and what should be done about it.

Insulin: The Hidden Driver of Metabolic Aging

Among the biomarkers that shape biological age, fasting insulin occupies an unusual position. It is one of the most metabolically significant signals the body produces, and one of the most consistently ignored in standard clinical practice.

The reason it gets ignored is that glucose and HbA1c have become the default markers for metabolic health assessment, and for most of the disease course they appear normal. This is precisely the problem. Blood glucose can remain within acceptable ranges for years, sometimes a decade or more, while the underlying physiology has already shifted toward insulin resistance. The body compensates for declining insulin sensitivity by producing more insulin, keeping glucose controlled at the cost of driving insulin higher. The compensation works, until it does not. And by the time glucose begins to rise and the standard markers finally signal a problem, the metabolic dysfunction has been present and accumulating damage for a long time.

Fasting insulin captures that earlier window. Research has shown that hyperinsulinemia can precede the development of type 2 diabetes by ten years or more, making it one of the earliest measurable indicators of metabolic disease progression available from a routine blood draw [6]. For anyone interested in early detection and intervention rather than late-stage disease management, that timing difference is not a minor technical detail. It is the difference between catching a problem when it is still highly responsive to intervention and catching it when it has already been compounding for years.

From a biological aging perspective, insulin's significance extends well beyond glucose regulation. Insulin is a central regulator of nutrient sensing, the signaling system that tells cells whether conditions favor growth or maintenance. Chronically elevated insulin drives persistent mTOR activation, suppresses autophagy, and impairs the metabolic flexibility that healthy cells depend on to shift efficiently between fuel sources [7]. Over time, this environment does not simply elevate disease risk. It accelerates the accumulation of metabolic stress across multiple organ systems simultaneously, producing the kind of systemic physiological drift that biological age models are designed to detect.

The original Bortz model did not include fasting insulin, a consequence of the biomarker set available within the UK Biobank dataset rather than a judgment about its relevance [2]. Recognizing this gap, Dr. Lanzilotti and Zsolt Szabo incorporated fasting insulin into updated implementations of the model specifically designed for clinical use within the Healthspan framework. The modification is more than incremental. By adding insulin, the model becomes sensitive to early metabolic dysfunction that glucose and HbA1c cannot yet see, shifting biological age estimation further upstream toward the phase when lifestyle modification, targeted supplementation, and clinical intervention have the greatest opportunity to redirect the trajectory. But insulin, as important as it is, does not operate in isolation. Understanding metabolic health fully requires seeing how it moves in relation to the broader set of markers that together define the state of the system. 

The Glucose Metabolism Model: Catching Metabolic Dysfunction Before Standard Panels Do

Insulin is a critical signal, but it does not tell the whole story on its own. Metabolic health is a system, and systems require more than one measurement point to understand properly.

This is the reasoning behind a more targeted glucose metabolism model developed within the Healthspan framework, one that integrates multiple biomarkers to build a complete picture of how the metabolic system is functioning rather than how any single component is performing. The model evaluates glucose, insulin, C-peptide, triglycerides, HDL, and HbA1c together, assessing insulin resistance, beta-cell function, and overall glucose regulation as an interconnected set of variables rather than independent values to be read in isolation.

The practical centerpiece of this approach is HOMA-IR, a composite metric that combines fasting glucose and fasting insulin into a validated estimate of insulin sensitivity. HOMA-IR serves as a practical proxy for the euglycemic clamp, the gold-standard laboratory method for measuring insulin resistance directly, but one that requires specialized equipment and controlled conditions that make it unavailable outside of research settings. By deriving a comparable signal from two routine markers, HOMA-IR brings meaningful insulin sensitivity assessment into standard clinical practice.

Equally important is what the model does with C-peptide. Where insulin reflects demand, C-peptide reflects supply, specifically the secretory output of the pancreatic beta cells responsible for producing insulin in the first place. Tracking both simultaneously allows the model to distinguish between two very different metabolic states that standard panels often conflate. A person in early insulin resistance may show elevated insulin and normal C-peptide, the body compensating successfully but under increasing strain. A person with later-stage metabolic failure may show declining insulin alongside falling C-peptide, as beta-cell function deteriorates and the compensatory capacity that had been holding glucose in check begins to fail. These are not the same clinical situation, and they call for different responses.

Standard laboratory panels identify metabolic disease after multiple regulatory systems have already deteriorated significantly. This model is designed to detect dysfunction earlier, when the physiology is still highly adaptable and intervention can redirect the trajectory rather than manage its consequences. Within the broader Healthspan framework, it functions as a natural complement to biological age estimation. Where the Bortz model captures the downstream systemic impact of metabolic dysfunction as reflected in biological age, the glucose metabolism model identifies the upstream drivers generating that impact, creating a more complete picture of both where a person stands and what is most likely responsible for putting them there.

Cystatin C: Rethinking Kidney Function

Creatinine has served as the standard marker of kidney function for decades, and for most populations it does the job adequately. But in the longevity and performance space, where resistance training, higher protein intake, and creatine supplementation are common, creatinine becomes a unreliable guide. All three of these factors elevate creatinine independently of kidney function, meaning individuals who are actively optimizing their health are precisely the ones most likely to receive a misleading signal from the marker their clinician is using to assess one of their most important organs.

Cystatin C solves this problem. Produced at a relatively constant rate by all nucleated cells in the body, it is largely unaffected by muscle mass, protein consumption, or creatine supplementation. This stability makes it a more accurate estimator of glomerular filtration rate, the measure of how efficiently the kidneys are clearing waste from the bloodstream, particularly in metabolically active and physically trained individuals [8]. Where creatinine reflects both kidney function and the muscle and dietary variables that can mimic kidney dysfunction, cystatin C reflects kidney function more cleanly.

Its value, however, extends beyond filtration accuracy. Subtle reductions in kidney function detected through cystatin C often appear before conventional thresholds for chronic kidney disease are reached, and they frequently travel alongside early changes in vascular health, endothelial function, and metabolic regulation. This makes cystatin C less a kidney-specific marker and more a sensitive indicator of systemic physiological stress, one that happens to be read in the kidney but reflects conditions that are developing more broadly.

This broader interpretive value becomes particularly apparent when cystatin C is read in relationship with other markers rather than in isolation. A modest elevation alongside rising insulin or triglycerides tells a different story than the same elevation alongside normal metabolic markers. The former may suggest early metabolic and vascular strain that is beginning to express itself across multiple systems simultaneously, even if each individual value remains within conventional normal ranges. This is precisely the kind of pattern that biological age models are designed to detect and that single-marker clinical panels routinely miss.

Understanding these relationships is what separates monitoring from optimization. Reporting individual values is the beginning of the process, not the end. Interpreting how cystatin C moves in relation to markers of metabolic health, lipid status, and inflammatory signaling is what allows a clinician or coach to identify what is actually driving early physiological drift and design interventions that address the underlying biology rather than the individual numbers. Improving insulin sensitivity, managing blood pressure, and reducing systemic inflammatory load can all influence kidney function over time, and cystatin C provides a sensitive readout of whether those interventions are working.

Within the Healthspan framework, integrating cystatin C into routine laboratory panels and biological age estimation does two things simultaneously. It improves the accuracy of kidney function assessment for a population where creatinine systematically misleads, and it adds a sensitive systems-level signal that enriches the broader picture of how different physiological domains are interacting. In this context, cystatin C is not simply a better kidney marker. It is a more honest window into the interconnected physiology that biological age models are ultimately trying to read. And reading that interconnected physiology accurately is only the first step. The more important question is what to do with it.

From Data to Action: Turning Biological Age into Strategy

A biological age score is a summary. Like any summary, its value depends entirely on what you do with it.

The number itself tells you where you are. It does not tell you why you are there, which systems are pulling that number in the wrong direction, or what would be most effective to change it. Getting from measurement to meaningful action requires moving beneath the composite score and into the underlying physiology that produced it.

This means abandoning single-marker thinking. Biological age is not shaped by any one biomarker in isolation. It emerges from the relationships between markers, how insulin moves alongside triglycerides, how cystatin C tracks with blood pressure and metabolic load, how liver enzymes interact with inflammatory signals and hematologic patterns. When these variables begin to drift together, even modestly and in ways that keep each individual value within conventional normal ranges, they can reveal coordinated physiological shifts that a standard panel reading would miss entirely. The signal is in the pattern, not the individual data points.

This is why interpretation is as important as measurement, and in many cases more difficult. A modest elevation in a single marker may carry limited significance on its own. The same elevation read in the context of related markers moving in the same direction is a different finding, one that can point toward early metabolic dysregulation, vascular strain, or declining organ resilience long before any single threshold has been crossed. Developing the clinical judgment to recognize these patterns, and to distinguish meaningful drift from normal biological variation, is what separates monitoring from optimization.

In practice, this means using the biological age model not as a label but as a map. The map identifies which physiological systems are contributing most to biological age acceleration, and those systems become the priority for intervention. For one person, the dominant driver may be insulin resistance and early metabolic dysfunction. For another, it may be liver stress, thyroid signaling, or early changes in kidney filtration. The model provides direction. The strategy targets the underlying biology.

What makes this approach genuinely powerful is its iterative nature. An intervention, whether through nutritional change, exercise programming, sleep optimization, or targeted clinical therapy, is implemented to shift specific biomarkers. Follow-up testing then provides objective feedback. Did the markers move? In which direction and by how much? Did related markers shift in parallel, suggesting the intervention is engaging the system more broadly, or did the primary marker improve while adjacent signals remained unchanged? These questions cannot be answered without measurement, and they cannot be asked intelligently without understanding the relationships between markers in the first place.

Over time, this creates the kind of feedback loop that makes biological age genuinely dynamic rather than static. Data informs action. Action reshapes physiology. Updated data refines the next step. The biological age estimate changes not because time has passed but because the underlying physiology has shifted, and the model is sensitive enough to register that shift and reflect it in its output.

The broader implication is straightforward but worth stating directly. Longevity is not optimized by a single test, a single intervention, or a single metric. It is optimized by the continuous, informed alignment of multiple physiological systems over time. Biological age models provide the structure for that process. Consistent, intelligent interpretation of the laboratory data underneath them is what drives it forward.

What the Number Cannot Tell You

Biological age models are powerful tools. They are not oracles.

Every model reviewed in this article, including the Bortz model, derives its estimates from statistical associations between biomarkers and mortality risk rather than from direct measurement of the biological processes driving aging. This distinction matters. The model is not reading aging itself. It is reading signals that correlate with aging outcomes across large populations and using those correlations to generate a probabilistic estimate of where an individual's physiology currently sits relative to those population-level risk curves. That is genuinely useful information. It is also inherently indirect, and it carries the uncertainty that comes with any probabilistic tool applied to an individual rather than a population.

Individual variability compounds this uncertainty in ways that are important to understand before acting on any single result. Biomarker levels are not fixed properties of a person's physiology. They fluctuate in response to genetics, environment, sleep quality, recent illness, hydration status, time of day, and the specific laboratory platform used to measure them. A biological age estimate taken during a period of acute stress, poor sleep, or illness will look different from one taken under stable conditions, not because biological age has changed dramatically but because the biomarkers feeding the model have shifted temporarily. This is not a failure of the model. It is a feature of physiology that any honest interpretation has to account for.

The practical implication is that a single biological age result should never be the primary basis for a clinical decision. The real value of these models emerges over time, when repeated measurements under consistent conditions begin to reveal patterns that a single data point cannot. Is biological age trending in the right direction following an intervention? Are specific marker clusters shifting in ways that suggest a particular physiological system is responding? These questions require longitudinal data to answer, and they are far more informative than any single score.

Within the Healthspan framework, this perspective is foundational rather than a disclaimer. The goal is not to produce a number and report it. It is to use that number as one input within a broader clinical picture, interpret it in the context of the marker relationships and domain-specific models that surround it, track it over time to distinguish meaningful physiological change from measurement noise, and use the pattern that emerges to guide interventions that target the underlying biology. Biological age is the starting point of that process, not its conclusion.

From Measurement to Momentum

For most of clinical history, a laboratory result was a diagnostic tool. It told you whether something had gone wrong. The framework described in this article asks a different question of the same data: not whether disease is present, but which direction physiology is heading, how quickly, and what can be done to change it before the trajectory becomes irreversible.

That shift in question changes everything that follows. It changes what gets measured, how results are interpreted, what counts as an actionable finding, and what success looks like over time. A biological age estimate is not a diagnosis. It is a directional signal, one that becomes most valuable not at the moment it is generated but in the months and years of measurement, interpretation, and intervention that follow.

That capacity is only valuable if it is acted upon. A biological age estimate sitting in a lab report, uninterpreted and unconnected to a clinical strategy, changes nothing. What changes things is the process that follows the measurement: identifying which physiological systems are driving the number, understanding how those systems interact, implementing interventions targeted at the underlying biology, and returning to the data to assess whether the physiology has moved. Done consistently, this transforms biological age from a static score into a dynamic feedback instrument, one that reflects not just where a person's physiology currently sits but whether the decisions they are making are shifting it in the right direction.

This is the deeper promise of the framework described in this article. Not simply more precise measurement, though precision matters. Not simply a better model, though the Bortz refinements are real and meaningful. But the integration of validated research tools with clinical partnership, longitudinal monitoring, and targeted intervention into a coherent approach that gives individuals genuine agency over the trajectory of their own aging.

The goal, ultimately, is not to know your biological age. It is to change it. And the evidence increasingly suggests that with the right measurements, the right interpretation, and the right strategy, that is exactly what becomes possible.