The entropic view of aging: from thermodynamics to biology

Abstract

A growing body of evidence suggests that aging is not simply the accumulation of damage, but a systemic progression of increasing entropy across biological scales. Synthesizing concepts from thermodynamics and information theory, we outline a tentative three-stage entropy model of the lifespan: order-building development, homeostatic adulthood, and disorder-dominant aging. Within this model, we attempt to reframe the established hallmarks of aging as interconnected nodes in an entropy-centered network and detail the multiscale manifestations of disorder from molecules to systems. To quantify this trajectory, we introduce a Multiscale Entropic Aging Index (MEAI) as a proof-of-concept conceptual framework, designed to integrate measurements of disorder across biological levels. This framework seeks to offer a potential unifying, quantitative language for aging research and suggests that entropy reduction could serve as a testable mechanism underlying interventions, laying a principled foundation for the future development of biomarkers and rejuvenative strategies.

Graphical abstract

Introduction

Entropy, first defined in 1865 by Rudolf Clausius as a thermodynamic property, quantifies the disorder or randomness in a system macroscopically [1]. The second law of thermodynamics states that isolated systems naturally evolve toward higher entropy, a state of increased chaos and energy dissipation. In Clausius’s work, the core formula describes the change in entropy (S), not an absolute value. For a reversible process, the infinitesimal change in entropy is given by Equation (1):

\begin{array}{}\text{(1)}& dS=\frac{\delta Q_{\text{rev}}}{T},\end{array}

(dS, the infinitesimal change in the entropy of the system; δQ_rev, the infinitesimal amount of heat transferred to the system in a reversible process; T, the absolute temperature of the system at the moment of heat transfer)

In open biological systems, such as cells or organisms, entropy is continuously exchanged with the environment. Metabolic processes supply free energy that sustains cellular order, while irreversible dissipation generates entropy. A net increase in entropy reflects a state in which energy dissipation outweighs the system’s capacity for maintenance and repair, a condition commonly associated with aging. For example, mitochondrial dysfunction in aged cells reduces ATP production efficiency, potentially increasing entropy production at the subcellular level [2].

In 1877, Ludwig Boltzmann’s kinetic theory redefined it microscopically [3] by introducing Equation (2):

\begin{array}{}\text{(2)}& S=k_B\ln W,\end{array}

(k_B, Boltzmann’s constant; ln, the natural logarithm; W, the number of microstates corresponding to a macrostate).

W represents the number of distinct molecular/cellular configurations (microstates) of a biological system under a given phenotypic state (macrostate). For example, the 3D chromatin conformations of a young cell are restricted to a small set of functional microstates (low W, low S), while aged cells exhibit constitutive heterochromatin decondensation and thousands of aberrant chromatin conformations (high W, high S), directly increasing Boltzmann entropy at the epigenomic scale.

In information theory, Claude Shannon later adapted entropy to measure uncertainty in data, where higher entropy corresponds to greater unpredictability [4], as expressed by Equation (3):

\begin{array}{}\text{(3)}& H(X)=-{\displaystyle {\sum }_{i=1}^{n}p(x_i){\log }_{2}p(x_i)},\end{array}

(H(X), Shannon entropy of the random variable X (measuring the degree of disorder in the system); n, the total number of possible states or outcomes of X; p(x_i), the probability of states or outcomes (e.g. methylated vs unmethylated status at a CpG site, or expression level of a gene) occurring).

Unlike thermodynamic entropy, Shannon entropy (i.e. information entropy) quantifies informational disorder in biological systems (e.g. DNA methylation states, gene expression profiles) [5, 6]. These concepts converge in biology: living systems resist entropy by harnessing energy to maintain order [7, 8], but aging represents the gradual failure of this resistance [9]. In this biological context, terms like “disorder” and “chaos” refer to the loss of functional and structural orderliness, which is the macroscopic manifestation of increasing entropy.

Biological order is epitomized during development and evolution, which organize matter into precise, low-entropy structures [10]. For example, embryonic development follows genetic and epigenetic blueprints to form specialized tissues, a process that is macroscopically entropy-reducing, with accumulating information content as cells differentiate into functionally distinct lineages. It is important to note that this progression toward organismal order can involve controlled stochasticity at molecular scales (e.g. in DNA methylation dynamics or chromatin remodeling [11]), which serves as essential “exploration” to reach the final low-entropy state. Evolution similarly optimizes traits for fitness, creating informational and functional complexity [12]. Conversely, aging is characterized by a systemic shift toward disorder across scales, reflecting the thermodynamic inevitability of entropy accumulation when the preservation of informational stability fails [13]. Here, we synthesize these contrasting dynamics to present aging through the lens of a lifelong entropy trajectory, suggesting that the accumulation of disorder might provide a unifying, quantifiable framework for understanding its mechanisms and guiding interventions.

The three-stage entropy dynamics model over the lifespan

Building upon the framework of orderliness, we propose a hypothesis describing the lifelong trajectory of biological entropy, which unfolds in three distinct phases: programmed entropy reduction (developmental stage), dynamic homeostasis (adult reproductive stage), and systemic entropy increase (post-reproductive aging stage) (Fig. 1).

The first phase, programmed entropy reduction, dominates development. Starting from a single-cell zygote, organisms undergo division, differentiation, and morphogenesis to form complex compartmentalized, polarized structures (macroscale entropy reduction). This process involves biological reprogramming at multiple levels. Molecularly, for example, the establishment of precise DNA methylation and histone modification landscapes and the reorganization of chromatin into specific 3D architectures write biological information into the genome, ensuring correct gene expression patterns while defining cell fates [11, 14]. This is a quintessential information-accumulating, entropy-reducing process, where information is written and disorder is dispelled to create a highly ordered state [12].

The second phase represents a state approaching dynamic homeostasis of entropy production and dissipation throughout the reproductive adulthood. At this stage, robust maintenance mechanisms, such as DNA repair, protein turnover, and immune surveillance, counteract perturbations to sustain relatively stable, low entropy levels, supporting successful reproduction [13, 15–17]. Notably, even in adulthood, low-level damage accumulation occurs (e.g. oxidative DNA lesions), but these are cleared before they can drive systemic entropy elevation, thus representing a net entropy-balancing process.

Finally, the phase of systemic entropy increase characterizes post-reproductive aging. Once the core reproductive mission is accomplished, the individual’s contribution to the population gene pool declines, and the force of natural selection weakens. The substantial energetic and resource costs of maintaining a low-entropy, high-informational-integrity state no longer confer a clear evolutionary advantage. Consequently, maintenance systems falter [15–18], and unregulated entropy accumulation becomes dominant, distinguishing this from isolated damage, as entropy can integrate diverse damage types into a systemic collapse of order (e.g. DNA damage + epigenetic drift + proteostasis loss drive a net increase in cellular microstates, elevating Boltzmann entropy and disrupting cell identity). This ultimately links the loss of order to phenotypic hallmarks of aging [13].

This three-stage trajectory reconciles programmed and entropy theories of aging: the post-reproductive decline of somatic maintenance programs serves as a temporal permissive condition for entropy accumulation, while entropy dynamics mechanistically describe how disorder propagates across scales once this homeostatic capacity wanes. In this view, the program does not actively drive aging but sets the stage for thermodynamic inevitability. This model further suggests the possible existence of an entropic tipping point, a threshold beyond which molecular disorder may become self-amplifying, potentially contributing to the nonlinear rise in morbidity and mortality observed in human aging curves. Such a tipping point could reflect the macroscopic manifestation of cumulative microstate expansion across scales, shaped by both programmed maintenance robustness and stochastic entropic insults. The population-level aging inflection point might thus be reframed as an echo of an underlying entropic phase transition, offering a conceptual target for interventions aimed at postponing this critical threshold. It is critical to recognize that entropy changes may not synchronize across all biological scales simultaneously. For example, during aging, molecular-scale entropy (e.g. in the (epi)genome) might accumulate first, eventually triggering a sharp rise in cell-, tissue- and system-scale entropy, marking functional collapse.

Entropy increase in aging: a multiscale perspective

Molecular and subcellular disorder

Aging drives a systemic increase in entropy across all biological scales (Fig. 2). Genomically, rising informational entropy parallels genomic instability, which encompasses DNA damage, mutations, telomere attrition, and retrotransposon reactivation, contributing to age-related diseases such as neurodegenerative disorders [19–24]. Epigenetically, entropy increases through altered DNA methylation, histone modification redistribution, and chromatin remodeling, leading to loss of compartmentalization and gene dysregulation [5, 20, 25–32]. For example, in Chinese individuals, epigenetic entropy, which is calculated as the Shannon entropy of CpG methylation states, shows a significant increase with age when comparing elder (> 70 years old) to younger adults (45–70 years old); whereas, long-lived individuals (> 90 years old) display a lower methylation entropy than elderly controls, which correlates with preserved heterochromatin structure and reduced age-related disease risk [28].

Transcriptionally, the leaky expression of developmentally restricted genes (e.g. placenta-specific genes (PSGs), PAPPA) and repetitive elements (e.g. endogenous retroviruses (ERVs)) in senescent cells represents a manifestation of elevated transcriptional entropy [26, 30, 31, 33–35]. The aging-driven disruption of gene expression pattern at the transcriptional level amplifies transcriptional noise or expression divergence, underscoring the loss of inherent transcriptional order [36–39]. Recently, a concept of network structural entropy derived from single-cell RNA sequencing data has been proposed. Studies applying this metric have demonstrated its increase in senescent cells, which correlates with the dysregulation of inherent cellular functions [40]. Moreover, single-cell transcriptome-based measures of cellular entropy not only correlate with senescence but also exhibit oncogenic potential, thereby linking transcriptional disorder to both proliferative arrest and cancer risk [41]. Yet, existing supervised approaches for detecting cellular senescence from single-cell transcriptomic data, while capable of generating cell-type-specific senescence scores, are limited by their dependence on annotated data and lack generalizability. A proposed universal framework overcomes this by quantifying single-cell transcriptional entropy through deviation from the transcriptomic manifold, capturing the breakdown of transcriptional coordination. This method identifies aging-affected cell types and distinguishes loss of expression precision from activated stress responses in high-entropy cells [42], offering a novel perspective for applying transcriptional entropy in aging assessment.

At the protein level, misfolding and aggregation, such as Aβ deposition observed in aged neuronal cells and associated with Alzheimer’s disease, lead to unregulated diversity of protein conformations and loss of functional specificity [43]. From a systemic perspective, this can be quantified as a reduction in the orderliness of the proteostatic network, or an increase in “proteostatic entropy”. Similarly, perturbations in nutrient sensing and metabolism disrupt the coordination and stability of metabolic networks, characterized by the dysregulation of key metabolic regulators, metabolites, enzymes, and even ions [43, 44]. For example, the recently defined ferro-aging pathway, driven by iron overload and lipid peroxidation, serves as a compelling molecular instantiation of the entropic principle, wherein stochastic oxidative damage to macromolecules manifests as a progressive, system-wide increase in disorder and functional decline [45]. This process reflects a decrease in the orderliness of metabolic systems, which can be conceptualized as an elevation of “metabolic entropy”. These molecular disorders together reflect the loss of informational and functional integrity within the cell.

Subcellular structures also lose their ordered architecture with aging (Fig. 2). For example, aged cells typically exhibit ruptured nuclear envelope and irregular nuclear morphology, as well as structural dysregulation in nuclear pores and nucleoli [43, 46–50]. Beyond the increasing chaos in the nucleus, a variety of organelles in the cytoplasm also display a rise in structural entropy. For example, the number of swollen or cristae-disorganized mitochondria often increases with aging of the cell [43, 51, 52]. Similarly, disrupted endoplasmic reticulum arrangement and abnormal lysosomal size and number also serve as potential indicators of increased structural entropy in organelles [20, 43, 53, 54]. Such structural disorder correlates with functional decline and loss of cellular homeostasis, underscoring the multiscale nature of entropic accumulation in aging.

Tissue and systemic disorder

Tissue-level entropy manifests as the loss of spatial order and functional specialization, suggesting escalating chaos in physiological, histological, cellular, and molecular features with aging (Fig. 2). For example, a large cohort study based on magnetic resonance imaging analysis reflecting brain activities revealed an increase in functional entropy with aging [55–57]. An entropy model based on electrocardiogram data has been reported to predict aging-related outcomes including fracture and mortality risks, which not only reflects physiological disorder at the organ level (i.e. heart) but also serves as a systemic indicator of homeostatic dysregulation at the organismal level [58]. Single-cell RNA sequencing of aging human lung tissues demonstrated a concurrent increase in somatic mutations, transcriptional perturbations, and shifts in cellular composition [59]. This convergence reveals a loss of inherent order across multiple levels and a corresponding rise in transcriptional entropy. A recent work introduced the concept of organizational structure entropy (OSE) for measuring the degree of disorder in tissue aging using spatial transcriptomic data and identified an increased OSE in aged tissues like hippocampus, spleen, and liver. Moreover, tissues with high OSE values preferentially show an enrichment of immune cells and a loss of cellular identity [60].

Consistent with the loss of identity, a gene expression profiling study across more than 40 tissue types revealed that aging is associated with a broad downregulation of pathways maintaining core cellular functions and a concurrent upregulation of genes related to epithelial–mesenchymal transition [61]. This phenomenon has been termed “mesenchymal drift”, which extends beyond epithelium to encompass endothelial cells, pericytes, and macrophages, leading to a collective “identity blurring”—essentially loss of cell type-specific informational identity [61]. These cells do not fully transdifferentiate but acquire superimposed mesenchymal traits, impairing their original functions while endowing them with new capabilities, such as modifying the extracellular matrix and altering tissue stiffness. Importantly, the pervasive mesenchymal drift signature is implicated in the pathogenesis of multiple age-related chronic diseases, including idiopathic pulmonary fibrosis, metabolic dysfunction-associated steatohepatitis, and Alzheimer’s disease [61]. Such drift can be viewed as a manifestation of informational loss (or an increase in entropy) and may constitute a fundamental mechanism driving organ dysfunction, aging, and chronic disease progression [62].

Apart from the infiltration of immune cells, aged tissues commonly exhibit an upregulation of inflammatory factors, a state referred to as the senescence-associated secretory phenotype (SASP), demonstrating pan-tissue inflammation during aging [43]. Such disorder is further manifested by structural alterations such as fibrosis and ectopic fat accumulation. Fibrosis, driven by fibroblast expansion and high extracellular matrix expression, disrupts tissue order in organs like heart, liver, and lungs. Similarly, ectopic fat deposition in non-adipose organs (e.g. skeletal muscle, liver, bone marrow) compromises tissue identity and function [43]. The breakdown of orderliness is further compounded by a decline in physiologically appropriate intercellular interactions. For instance, disruption of intercellular tight junctions, as recently reported in the aged primate small intestine, impairs epithelial barrier integrity and contributes to tissue disorganization [63]. Moreover, noise and chaos in cell-to-cell communication emerge as non-specific, destructive signals (e.g. the aforementioned pro-inflammatory signals) are indiscriminately released, disrupting the signaling balance of the local microenvironment [64]. For example, recent studies have uncovered that aged tissues accumulate immunoglobulins, especially IgG, which drives a senescence-associated phenotype and fuels multi‑organ inflammation [60, 65]. At a finer scale, cellular senescence within aged tissues introduces multidimensional molecular disturbances across the genome, epigenome, transcriptome, proteome, and metabolome, amplifying tissue-level disorder [16, 43, 60, 66–68].

Systemically, as beneficial hormonal cues wane and disruptive signaling emerges, the integrative capacity of vital systems, including the nervous, endocrine, and immune systems, is impaired, leading to decreased homeostasis and increased vulnerability to age-related diseases [69]. This shift from stability to disorder indicates a rise in organismal entropy, though quantitative models are still needed to formally establish this link. The most direct physiological manifestation of this transition is a marked decline in organismal resilience and a corresponding increase in fragility [70–77]. Here, “resilience” refers to the ability of a system to restore its original steady state after a disturbance (such as infection, trauma, or thermal stress), while “fragility” denotes the tendency of a system to undergo functional collapse under perturbation. During aging, multi-scale increases in entropy, from molecular to cellular, tissue, and systemic levels, progressively erode the redundancy and coordination necessary for sustaining complex functions. Consequently, when confronted with stressors, the aged organism not only recovers more slowly (loss of resilience) but is also more likely to cross a critical threshold into irreversible functional decline or disease (increased fragility). This reflects the approach of the system as a whole toward a critical threshold of disorder, providing an entropy-based dynamical explanation for the steep rise in age-related morbidity and mortality. This systemic loss of homeostatic order has been empirically captured in a human cohort, where a rising dyshomeostasis index (DHI), which integrates multi-omics and clinical parameters, quantitatively reflects the entropic principle of increasing disorder across multiple physiological systems [78].

Entropy as a unifying framework for aging research

How the entropic framework refines and unifies the hallmarks of aging

The entropic framework is more than a relabeling of known hallmarks; it provides a unifying, quantitative lens that refines our understanding. The current hallmarks framework [79–81], while highly influential, catalogs an expanding list of features but leaves their interrelationships largely unspecified, nor does it identify which might operate most upstream. The entropy framework offers a complementary perspective: it posits that regardless of the initial insult, if it ultimately increases the number of accessible system microstates and degrades informational capacity, it contributes to aging. Conversely, if a given type of damage does not entail such entropic increase, it may not represent a core driver of the aging process, a distinction that is, in principle, empirically testable. It further suggests that diverse age-related changes are interconnected manifestations of a single thermodynamic trend, i.e. loss of structural and functional information and order. This perspective shifts the focus from cataloging independent types of damage to investigating how entropy propagates and amplifies across scales, potentially leading to systemic failure at critical thresholds. The core value of the entropic perspective lies in its ability to recast established aging hallmarks as interconnected nodes in a network driving systemic disorder, prompting novel, quantifiable questions. Importantly, it generates the testable prediction that effective aging interventions should produce a measurable decrease in entropy metrics at one or more scales, providing a universal benchmark for efficacy beyond specific physiological outcomes.

Assessment: quantitative entropy-based biomarkers

Biological entropy is commonly quantified using Shannon’s formula. This measure facilitates cross-scale disorder comparison from molecular states to system-level function. Critically, it allows different types of disorder (e.g. epigenetic drift, transcriptional perturbation) to be placed on a common quantitative scale. Increased entropy at genomic, epigenetic, transcriptional, and functional levels has been demonstrated using this approach, showing predictive potential for cellular and tissue aging that can extend beyond conventional metrics [5, 19, 26, 41, 55, 60, 82]. For instance, methylation entropy has been shown to be lower in healthy long-lived individuals and can distinguish health status independently of chronological age [28]. Similarly, transcriptional entropy calculated from single-cell RNA-seq data can identify pre-malignant cell states not captured by traditional markers [41].

While individual entropy metrics offer powerful biomarkers, translating the entropic theory of aging into a unified, practical index faces two critical challenges. First, the non-monotonicity of entropy change over the lifespan presents a conceptual hurdle: if the central hypothesis holds, that entropy decreases during development, reaches a nadir in young adulthood, and increases thereafter, then a given entropy value could correspond to two distinct biological ages (one in development, one in aging), making it an ambiguous standalone measure of aging. Second, there is the multiscale complexity of biological systems. Aging is a systemic, heterogeneous, and nonlinear process manifesting across distinct dimensions (molecular, cellular, tissue, etc.) [16, 83, 84], each requiring specific quantification methods. A valuable aging index must therefore integrate this complexity into a coherent single metric.

To explore a potential solution to these limitations, we here propose a conceptual framework to measure entropy across multiple biological scales, from molecules to tissues, and combine them into a hypothetical composite index. This approach circumvents the non-monotonicity problem in theory by using the young adult homeostatic state as a reference point, focusing on the entropic increase that characterizes the aging phase. This leads to the formulation of a Multiscale Entropic Aging Index (MEAI) as a theoretical construct, given by Equation (4):

\begin{array}{}\text{(4)}& \text{MEAI}={\displaystyle {\sum }_{i=1}^M{w}_{\text{i}}\frac{H_i^{(\text{current})}-{\mu }_{H_i^{(\text{ref})}}}{{\sigma }_{H_i^{(\text{ref})}}}},\end{array}

Here, M represents the number of biological scales selected. The weight w_i (where ∑w_i = 1) reflects the relative contribution of each scale’s disorder to systemic aging. H_i^{(\text{current})} represents the entropy (e.g. genetic entropy, epigenetic entropy, transcriptional entropy, proteomic entropy, metabolic entropy, cell type entropy, tissue organization entropy, functional entropy) at biological scale i for the individual being assessed. {\mu }_{H_i^{(\text{ref})}} represents the mean entropy at biological scale i in a large, well-characterized young healthy reference population (where entropy reaches a stable nadir post-development). {\sigma }_{H_i^{(\text{ref})}} represents the standard deviation of entropy at biological scale i in the reference population. Notably, since MEAI is designed to specifically quantify aging-related entropic increase, the formula holds under the premise that H_i^{(\text{current})}>{\mu }_{H_i^{(\text{ref})}}. The MEAI offers a theoretical basis for quantifying relative entropic increase across scales, representing a hypothetical single metric that would, in principle, increase with accumulated disorder, though its practical implementation as a unified index remains pending.

As a proof-of-concept, MEAI requires extensive implementation, optimization, and empirical validation to address unmet practical challenges. First, the assignment of w_i is non-trivial and may be informed by data-driven approaches (e.g. feature importance from machine learning models) or expert consensus, yet it must ultimately be refined through testing on large-scale longitudinal datasets. Second, ensuring comparability of entropy values across disparate biological scales necessitates a consistent computational framework (e.g. standardized state/bin definitions for Shannon entropy calculation). Third, defining the robust reference values ({\mu }_{H_i^{(\text{ref})}} and {\sigma }_{H_i^{(\text{ref})}}) is critical. This requires establishing a baseline from a clearly characterized cohort of healthy young individuals, with mean/median entropy from large cohorts (strict health criteria) and stratified values (e.g. by sex) to boost population applicability. Fourth, linear weighted summation may oversimplify nonlinear synergies between multi-scale disorder, advanced models (e.g. graph neural networks, multi-scale factor analysis) may be needed to capture how molecular entropy amplifies tissue and systemic dysfunction. Fifth, computational feasibility remains a hurdle: some entropy metrics (e.g. tissue structural entropy from spatial transcriptomics, organ functional entropy from imaging) lack standardized pipelines, requiring reproducible tool development for cross-scale quantification. Given its current theoretical nature and unresolved practical challenges (e.g. cross-scale entropy comparability, weight calibration, individual-level stability, and nonlinear integration), the predictive performance of MEAI and its optimization in diverse populations remain to be explored and established through large-scale future empirical studies. Initial proof-of-concept studies for MEAI should prioritize integrating two to three well-validated entropy metrics (e.g. DNA methylation entropy + single-cell transcriptional entropy), testing whether the composite index outperforms single metrics in predicting aging phenotypes or intervention efficacy, this incremental validation will refine the framework before expanding to complex multi-scale integration.

Notably, MEAI’s reliance on Shannon entropy introduces a fundamental methodological consideration. As a distribution-based metric, Shannon entropy requires a predefined probability distribution across discrete states or bins, an inherent characteristic that complicates assigning meaningful entropy values to individual samples without reference to a population baseline. This limitation underscores the broader challenge of personalized aging assessment, where single-individual measurements need to be interpreted against normative data. In light of these considerations, recent methodological discussions have highlighted complementary approaches such as the Mahalanobis distance, which quantifies how far an individual’s multi-dimensional biomarker profile deviates from a healthy reference population [58, 85]. Unlike Shannon entropy, this metric can be calculated from a single observation across multiple variables and may offer a more direct measure of physiological divergence, one that could complement the cross-scale entropy integration envisioned by MEAI, once its practical limitations are addressed in future work.

Collectively, translating the entropic aging framework and realizing the potential of MEAI as a unified index from theory to practice hinges on several critical, yet unmet steps: building standardized population cohorts to capture normative aging trajectories, standardizing entropy calculations across scales, establishing robust reference values, and developing integrated pipelines to synthesize multi‑scale measures into a coherent “entropic age” index. Future advances will likely rely on machine‑learning models trained on large multi‑omics cohorts, potentially hybridizing entropy‑based and distance‑based measures to capture both systemic disorder and individual physiological divergence, thereby paving the way for a next generation of mechanism‑informed biomarkers.

Intervention: targeting entropy reduction with mechanistic specificity

Rejuvenative interventions that target aging-related biological processes leverage natural “negative entropy”-associated mechanisms to restore functional and structural order (Fig. 1). This framework allows us to contextualize their actions in entropic terms: interventions improve core biological functions, with reduced systemic disorder (entropy) emerging as a downstream readout of these beneficial changes. For example, caloric restriction may lower systemic entropy by enhancing metabolic efficiency and selective autophagy, specifically removing high-entropy, dysfunctional cellular components like damaged mitochondria [86–88]. Exercise, while transiently boosting energy expenditure, drives long-term improvements in biological order by preventing DNA damage and telomere shortening, mitigating DNA methylation drift, and restoring heterochromatin stability [9, 89, 90]; such enhancements in genomic and epigenetic integrity potentially reflect the reduced genomic and epigenetic entropy.

Recent studies have demonstrated that inactivation of a single factor (e.g. SIRT2, SIRT5, and GRHL2) or activation of others (e.g. KAT7, ERVs, and LINE-1) is sufficient to initiate the programmed aging process, while targeted intervention of these factors can achieve rejuvenation through reprogramming by resetting core transcriptional programs and restoring information homeostasis and order [34, 91–95]. This suggests that maintaining the expression control capacity of individual factors enables the system to preserve its overall negentropic capacity. Similarly, rejuvenation strategies, including administering geroprotective compounds (e.g. metformin, vitamin C, and uridine) or delivering youth-promoting genes (e.g. CLOCK, SOX5, and Yamanaka factors), can counteract aging by reestablishing genetic and epigenetic order [96–107]. Such restoration of molecular organization aligns with the conceptual framework of reduced biological entropy as a downstream outcome. Furthermore, the strategy of intermittently introducing Yamanaka genes has been shown to alleviate aging phenotypes through reversing mesenchymal drift, which points to the re-establishment of cellular identity order [61]. Other interventions such as senolytics directly remove high-entropy senescent cells, a major source of disorder in aged tissues, while stem cell transplantation introduces low-entropy, information-rich elements to reset local tissue organization [108–110]. These actions improve tissue structure and function, with reduced multi-scale entropy serving as a unifying readout of their biological efficacy. Therefore, a central, testable prediction of this framework is that diverse aging interventions, by improving distinct biological mechanisms, should converge on reducing measurable entropy at multiple scales, offering a potential universal biomarker for their deep biological effect.

Conclusion and future perspective

In summary, aging can be usefully viewed as a systemic process of entropic accumulation, spanning molecular disorder, cellular identity loss, and tissue dysfunction. As a working hypothesis, this framework unifies aging hallmarks by explaining them as interconnected contributors to a common trend toward rising disorder, with entropy metrics emerging as versatile biomarkers, ranging from (epi)genomic and transcriptional entropy for predicting cellular senescence, to entropy relying on cellular identity loss for quantifying tissue degeneration, and further to imaging-derived brain entropy for assessing aging and Alzheimer’s disease, as well as electrocardiogram-based entropy for forecasting aging-related adverse outcomes. Moving beyond framing entropy as primarily a metaphorical descriptor of aging, this framework seeks to advance a paradigm shift from philosophical analogy toward a rigorous operational scientific construct: by establishing multiscale operational definitions of biological entropy, from epigenetic methylation entropy to tissue structural entropy, and introducing MEAI as a relative metric that circumvents the non-monotonicity problem, it proposes to define entropy as a quantifiable state variable that could potentially unify programmed and stochastic views of aging within a common conceptual coordinate system.

This entropy-based view is further reinforced by its convergence with the Information Theory of Aging (ITOA), which posits the loss of epigenetic information as a molecular driver of aging [111]. The two perspectives are mutually illuminating: ITOA provides a mechanistic basis for “molecular entropy increase”, wherein noise-induced epigenetic disorder fuels cellular identity loss and expression chaos, mirroring multi-scale entropic increase. Both frame aging as systemic disorder amplification, from molecular to systemic function, and align on a dynamic trajectory across the lifespan: development (epigenetic writing/entropy decrease), adulthood (information homeostasis/entropy balance), and aging (information loss/entropy increase). Critically, ITOA translates the goal of “reducing disorder” into actionable strategies like partial reprogramming and enhanced DNA repair, aiming to lower molecular entropy by restoring youthful epigenetic landscapes. While ITOA focuses on epigenetic information as a key substrate, the present framework attempts to generalize “information” to encompass order across multiple biological scales, from epigenetic to transcriptional, proteostatic, metabolic, and tissue structural entropy. In this view, epigenetic entropy is but one manifestation of a broader systemic trend; the framework thus may help address how disorder propagates from molecules to organismal function, offering a complementary systems-level perspective that is consistent with ITOA while potentially extending its conceptual reach.

At present, no empirical evidence supports the linear additivity of entropy across scales; thus MEAI remains a conceptual scaffold rather than a ready‑to‑use metric. Looking forward, quantifying entropy via conceptual frameworks like MEAI, anchored to young adult reference entropy and requiring future calibration of data-driven weights, complemented by Mahalanobis distance for personalized assessment, offers potential for yielding cross-scale aging biomarkers with unique integrative potential in the future, while entropy-reversing interventions hold great promise for extending healthspan. Realizing this vision requires a concerted effort to develop standardized, cross-scale bio-entropy protocols, build large-scale “entropic phenotype” databases, and validate such framework in large longitudinal studies to address its current practical limitations. A speculative hypothesis worth testing is whether organ functional collapse might consistently follow a simultaneous, sharp rise in entropy across multiple scales (an “entropic critical point”). Interdisciplinary collaboration spanning thermodynamics, omics, clinical medicine, and artificial intelligence is essential to decode aging’s “entropic signature” and transition from managing late-life diseases to targeting the root disorder of the aging process itself.

References

[1] Clausius R. Ueber verschiedene für die Anwendung bequeme Formen der Hauptgleichungen der mechanischen Wärmetheorie. Ann Phys 1865;201:353.

[2] Poljšak B, Milisav I. Decreasing intracellular entropy by increasing mitochondrial efficiency and reducing ROS formation—the effect on the ageing process and age-related damage. Int J Mol Sci 2024;25:6321.

[3] Gimbel S. It ain’t necessarily so: Ludwig Boltzmann’s Darwinian notion of entropy. Entropy (Basel) 2024;26:238.

[4] Saraiva P. On Shannon entropy and its applications. Kuwait J Sci 2023;50:194-9.

[5] Vaidya H, Jeong HS, Keith K et al. DNA methylation entropy as a measure of stem cell replication and aging. Genome Biol 2023;24:27.

[6] García-Nieto PE, Wang B, Fraser HB. Transcriptome diversity is a systematic source of variation in RNA-sequencing data. PLoS Comput Biol 2022;18:e1009939.

[7] Phillips R. Schrödinger’s what is life? at 75. Cell Syst 2021;12:465-76.

[8] Skinner DJ, Dunkel J. Improved bounds on entropy production in living systems. Proc Natl Acad Sci USA 2021;118:e2024300118.

[9] Wang Z. The entropy perspective on human illness and aging. Engineering 2022;9:22-6.

[10] Yoram S. How organisms decrease their entropy. Biosystems 2025;257:105592.

[11] Burton A, Torres-Padilla ME. Epigenome dynamics in early mammalian embryogenesis. Nat Rev Genet 2025;26:587-603.

[12] Paul CWD, Elisabeth R, Jack AT. Self-organization and entropy reduction in a living cell. Biosystems 2013;111:1-10.

[13] Andres K. Dissipative scaling of development and aging in multicellular organisms. Biosystems 2024;237:105157.

[14] Wilkinson AL, Zorzan I, Rugg-Gunn PJ. Epigenetic regulation of early human embryo development. Cell Stem Cell 2023;30:1569-84.

[15] Yamada M, Udono MU, Hori M et al. Aged human skin removes UVB-induced pyrimidine dimers from the epidermis more slowly than younger adult skin in vivo. Arch Dermatol Res 2006;297:294-302.

[16] Ding Y, Zuo Y, Zhang B et al. Comprehensive human proteome profiles across a 50-year lifespan reveal aging trajectories and signatures. Cell 2025;188:5763-84.e26.

[17] Wang Y, Li R, Tong R et al. Integrating single-cell RNA and T cell/B cell receptor sequencing with mass cytometry reveals dynamic trajectories of human peripheral immune cells from birth to old age. Nat Immunol 2025;26:308-22.

[18] Tuyéras R et al. Cellular intelligence: dynamic specialization through non-equilibrium multi-scale compartmentalization. bioRxiv, 2021.2006.2025.449951, 2021, https://doi.org/10.1101/2021.06.25.449951, preprint: not peer reviewed.

[19] Riggs JE. Aging, increasing genomic entropy, and neurodegenerative disease. Neurol Clin 1998;16:757-70.

[20] Wu Z, Qu J, Liu GH. Roles of chromatin and genome instability in cellular senescence and their relevance to ageing and related diseases. Nat Rev Mol Cell Biol 2024;25:979-1000.

[21] Li X, Yu H, Li D et al. LINE-1 transposable element renaissance in aging and age-related diseases. Ageing Res Rev 2024;100:102440.

[22] Zhang H, Li J, Yu Y et al. Nuclear lamina erosion-induced resurrection of endogenous retroviruses underlies neuronal aging. Cell Rep 2023;42:112593.

[23] De Cecco M, Ito T, Petrashen AP et al. L1 drives IFN in senescent cells and promotes age-associated inflammation. Nature 2019;566:73-8.

[24] Simon M, Van Meter M, Ablaeva J et al. LINE1 derepression in aged wild-type and SIRT6-deficient mice drives inflammation. Cell Metab 2019;29:871-85.e5.

[25] Hannum G, Guinney J, Zhao L et al. Genome-wide methylation profiles reveal quantitative views of human aging rates. Mol Cell 2013;49:359-67.

[26] Liu Z, Ji Q, Ren J et al. Large-scale chromatin reorganization reactivates placenta-specific genes that drive cellular aging. Dev Cell 2022;57:1347-68.e12.

[27] Yang J-H, Hayano M, Griffin PT et al. Loss of epigenetic information as a cause of mammalian aging. Cell 2023;186:305-26.e27.

[28] Wang H-T, Xiao F-H, Gao Z-L et al. Methylation entropy landscape of chinese long-lived individuals reveals lower epigenetic noise related to human healthy aging. Aging Cell 2024;23:e14163.

[29] Jenkinson G, Pujadas E, Goutsias J et al. Potential energy landscapes identify the information-theoretic nature of the epigenome. Nat Genet 2017;49:719-29.

[30] Liu Z, Belmonte JCI, Zhang W et al. Deciphering aging at three-dimensional genomic resolution. Cell Insight 2022;1:100034.

[31] Zhao D, Chen S. Failures at every level: breakdown of the epigenetic machinery of aging. Life Med 2022;1:81-3.

[32] Zhang W, Li J, Suzuki K et al. A Werner syndrome stem cell model unveils heterochromatin alterations as a driver of human aging. Science 2015;348:1160-3.

[33] Bi S, Jiang X, Ji Q et al. The sirtuin-associated human senescence program converges on the activation of placenta-specific gene PAPPA. Dev Cell 2024;59:991-1009.e12.

[34] Liu X, Liu Z, Wu Z et al. Resurrection of endogenous retroviruses during aging reinforces senescence. Cell 2023;186:287-304.e26.

[35] Zhou S, Liu L, Lu X. Endogenous retroviruses make aging go viral. Life Med 2023;2:lnad001.

[36] Zhang W, Zhang S, Yan P et al. A single-cell transcriptomic landscape of primate arterial aging. Nat Commun 2020;11:2202.

[37] Zhang H, Li J, Ren J et al. Single-nucleus transcriptomic landscape of primate hippocampal aging. Protein Cell 2021;12:695-716.

[38] Wang S, Zheng Y, Li J et al. Single-cell transcriptomic atlas of primate ovarian aging. Cell 2020;180:585-600.e19.

[39] Wu Z, Lu M, Liu D et al. m6A epitranscriptomic regulation of tissue homeostasis during primate aging. Nat Aging 2023;3:705-21.

[40] Liu Z, Lin H, Li X et al. The network structural entropy for single-cell RNA sequencing data during skin aging. Brief Bioinform 2024;26:bbae698.

[41] Bai H, Liu X, Lin M et al. Progressive senescence programs induce intrinsic vulnerability to aging-related female breast cancer. Nat Commun 2024;15:5154.

[42] Yang Y et al. A manifold-based measure of transcriptional entropy for quantifying aging in single cells. bioRxiv, 2026.2001.2024.701460, 2026, https://doi.org/10.64898/2026.01.24.701460, preprint: not peer reviewed.

[43] Wu ZM, Qu J, Zhang WQ et al. Aging Biomarker Consortium. Biomarkers of ageing of humans and non-human primates. Nat Rev Mol Cell Biol 2025;26:826-47.

[44] Wiley CD, Campisi J. The metabolic roots of senescence: mechanisms and opportunities for intervention. Nat Metab 2021;3:1290-301.

[45] Liu L, Zheng Z, You W et al. Vitamin C inhibits ACSL4 to alleviate ferro-aging in primates. Cell Metab 2026;38:673-93.e17.

[46] Pathak RU, Soujanya M, Mishra RK. Deterioration of nuclear morphology and architecture: a hallmark of senescence and aging. Ageing Res Rev 2021;67:101264.

[47] Heckenbach I, Mkrtchyan GV, Ezra MB et al. Nuclear morphology is a deep learning biomarker of cellular senescence. Nat Aging 2022;2:742-55.

[48] Rempel IL, Steen A, Veenhoff LM. Poor old pores—the challenge of making and maintaining nuclear pore complexes in aging. FEBS J 2020;287:1058-75.

[49] Gutierrez JI, Tyler JK. A mortality timer based on nucleolar size triggers nucleolar integrity loss and catastrophic genomic instability. Nat Aging 2024;4:1782-93.

[50] Li H-Y, Wang M, Jiang X et al. CRISPR screening uncovers nucleolar RPL22 as a heterochromatin destabilizer and senescence driver. Nucleic Acids Res 2024;52:11481-99.

[51] Liu Y, Han M, Li X et al. Age-related changes in the mitochondria of human mural granulosa cells. Hum Reprod 2017;32:2465-73.

[52] He Y, Ji Q, Wu Z et al. 4E-BP1 counteracts human mesenchymal stem cell senescence via maintaining mitochondrial homeostasis. Protein Cell 2023;14:202-16.

[53] Wang S, Hu B, Ding Z et al. ATF6 safeguards organelle homeostasis and cellular aging in human mesenchymal stem cells. Cell Discov 2018;4:2.

[54] Tan JX, Finkel T. Lysosomes in senescence and aging. EMBO Rep 2023;24:e57265.

[55] Yao Y, Lu WL, Xu B et al. The increase of the functional entropy of the human brain with age. Sci Rep 2013;3:2853.

[56] Wang Z; Alzheimer’s Disease Neuroimaging Initiative. Brain entropy mapping in healthy aging and Alzheimer’s disease. Front Aging Neurosci 2020;12:596122.

[57] Lin C, Lee S-H, Huang C-M et al. Increased brain entropy of resting-state fMRI mediates the relationship between depression severity and mental health-related quality of life in late-life depressed elderly. J Affect Disord 2019;250:270-7.

[58] Hong N, Cho SW, Kim J et al. Entropy, assessed by homeostatic dysregulation on electrocardiograms predicts fracture and mortality. Aging Cell 2025;24:e70227.

[59] De Man R, McDonough JE, Adams TS et al. Single-cell atlas of human lung aging identifies cell type dyssynchrony and increased transcriptional entropy. Nat Commun 2026;17:2095.

[60] Ma S, Ji Z, Zhang B et al. Spatial transcriptomic landscape unveils immunoglobin-associated senescence as a hallmark of aging. Cell 2024;187:7025-44.e34.

[61] Lu JY, Tu WB, Li R et al. Prevalent mesenchymal drift in aging and disease is reversed by partial reprogramming. Cell 2025;188:5895-911.e17.

[62] Lu JY et al. Mesenchymal drift: a convergent framework for the hallmarks of aging. Cell 2026;189:3184-213.

[63] Li J, Lu X, Tong T et al. Single-nucleus resolution atlas of primate small intestinal aging reveals NCoR1 decline as a conserved feature that is reversed by metformin. Nat Aging 2026.

[64] Fafián-Labora JA, O’Loghlen A. Classical and nonclassical intercellular communication in senescence and ageing. Trends Cell Biol 2020;30:628-39.

[65] Yu L, Wan Q, Liu Q et al. IgG is an aging factor that drives adipose tissue fibrosis and metabolic decline. Cell Metab 2024;36:793-807.e5.

[66] Tang W, Wang K, Feng Y et al. Exploration of the mechanism and therapy of ovarian aging by targeting cellular senescence. Life Med 2025;4:lnaf004.

[67] Cai Y, Song W, Li J et al. The landscape of aging. Sci China Life Sci 2022;65:2354-454.

[68] Bao H et al. Biomarkers of aging. Sci China Life Sci 2023; 66:893-1066.

[69] Straub RH, Cutolo M, Zietz B et al. The process of aging changes the interplay of the immune, endocrine and nervous systems. Mech Ageing Dev 2001;122:1591-611.

[70] Cai Y, Xiong M, Xin Z et al. Decoding aging-dependent regenerative decline across tissues at single-cell resolution. Cell Stem Cell 2023;30:1674-91.e8.

[71] Ukraintseva S, Arbeev K, Duan M et al. Decline in biological resilience as key manifestation of aging: potential mechanisms and role in health and longevity. Mech Ageing Dev 2021;194:111418.

[72] Sierra F, Perez V. Loss of molecular resilience as the ultimate outcome of aging biology. Geroscience 2026;48:217-23.

[73] Howlett SE, Rutenberg AD, Rockwood K. The degree of frailty as a translational measure of health in aging. Nat Aging 2021;1:651-65.

[74] Woo J. Frailty, successful aging, resilience, and intrinsic capacity: a cross-disciplinary discourse of the aging process. Curr Geriatr Rep 2019;8:67-71.

[75] Sanchez KK, McCarville JL, Stengel SJ et al. Disease tolerance and infection pathogenesis age-related tradeoffs in mice. Nature 2026;650:727-35.

[76] Delval L, Hantute-Ghesquier A, Sencio V et al. Removal of senescent cells reduces the viral load and attenuates pulmonary and systemic inflammation in SARS-CoV-2-infected, aged hamsters. Nat Aging 2023;3:829-45.

[77] Jiang Z, Pan W, Chen Y et al. ApoD mediates age-associated increase in vulnerability to influenza virus infection. Proc Natl Acad Sci USA 2025;122:e2423973122.

[78] Li J, Jiang B, Zhang W et al. Multimodal clocks of human aging. Cell 2026;189:1-23.

[79] Lopez-Otin C, Blasco MA, Partridge L et al. The hallmarks of aging. Cell 2013;153:1194-217.

[80] López-Otín C, Blasco MA, Partridge L et al. Hallmarks of aging: an expanding universe. Cell 2023;186:243-78.

[81] Kroemer G, Maier AB, Cuervo AM et al. From geroscience to precision geromedicine: understanding and managing aging. Cell 2025;188:2043-62.

[82] Chan J, Rubbi L, Pellegrini M. DNA methylation entropy is a biomarker for aging. Aging (Albany NY) 2025;17:685-98.

[83] Shen X, Wang C, Zhou X et al. Nonlinear dynamics of multi-omics profiles during human aging. Nat Aging 2024;4:1619-34.

[84] Li G-H, Li Y-H, Yu Q et al. Unraveling the metabolic heterogeneity and commonality in senescent cells using systems modeling. Life Med 2025;4:lnaf003.

[85] Cummings SR, Hong N, Cohen AA. Entropy and human aging. Aging Cell 2025;24:e70292.

[86] Redman LM, Smith SR, Burton JH et al. Metabolic slowing and reduced oxidative damage with sustained caloric restriction support the rate of living and oxidative damage theories of aging. Cell Metab 2018;27:805-15.e4.

[87] Bagherniya M, Butler AE, Barreto GE et al. The effect of fasting or calorie restriction on autophagy induction: a review of the literature. Ageing Res Rev 2018;47:183-97.

[88] Ma S, Sun S, Geng L et al. Caloric restriction reprograms the single-cell transcriptional landscape of Rattus norvegicus aging. Cell 2020;180:984-1001.e22.

[89] Sellami M, Bragazzi N, Prince MS et al. Regular, intense exercise training as a healthy aging lifestyle strategy: preventing DNA damage, telomere shortening and adverse DNA methylation changes over a lifetime. Front Genet 2021;12:652497.

[90] Geng L, Ping J, Wu R et al. Systematic profiling reveals betaine as an exercise mimetic for geroprotection. Cell 2025;188:5403-25.e33.

[91] Ye Y, Yang K, Liu H et al. SIRT2 counteracts primate cardiac aging via deacetylation of STAT3 that silences CDKN2B. Nat Aging 2023;3:1269-87.

[92] Zhao Q, Jing Y, Jiang X et al. SIRT5 safeguards against primate skeletal muscle ageing via desuccinylation of TBK1. Nat Metab 2025;7:556-73.

[93] Sun G, He Z, Lv D et al. Reprogramming the GRHL2-CDK19 axis by gene therapy alleviates prostate aging. Nat Aging 2026;6:252-69.

[94] Wang W, Zheng Y, Sun S et al. A genome-wide CRISPR-based screen identifies KAT7 as a driver of cellular senescence. Sci Transl Med 2021;13:eabd2655.

[95] Yang C, Du H, Liu S et al. Targeting age-related LINE-1 activation alleviates cardiac aging. Nat Aging 2026;6:414-29.

[96] Yang Y, Lu X, Liu N et al. Metformin decelerates aging clock in male monkeys. Cell 2024;187:6358-78.e29.

[97] Li Y, Zhang W, Chang L et al. Vitamin C alleviates aging defects in a stem cell model for werner syndrome. Protein Cell 2016;7:478-88.

[98] Sun S, Li J, Wang S et al. CHIT1-positive microglia drive motor neuron ageing in the primate spinal cord. Nature 2023;624:611-20.

[99] Liu Z, Li W, Geng L et al. Cross-species metabolomic analysis identifies uridine as a potent regeneration promoting factor. Cell Discov 2022;8:6.

[100] Kim H, Yu MR, Lee H et al. Metformin inhibits chronic kidney disease-induced DNA damage and senescence of mesenchymal stem cells. Aging Cell 2021;20:e13317.

[101] Ocampo A, Reddy P, Martinez-Redondo P et al. In vivo amelioration of age-associated hallmarks by partial reprogramming. Cell 2016;167:1719-33.e2.

[102] Jing Y, Jiang X, Ji Q et al. Genome-wide CRISPR activation screening in senescent cells reveals SOX5 as a driver and therapeutic target of rejuvenation. Cell Stem Cell 2023;30:1452-71.e10.

[103] Chen Y, Xu X, Chen Z et al. Circadian factors CLOCK and BMAL1 promote nonhomologous end joining and antagonize cellular senescence. Life Med 2024;3:lnae006.

[104] Liang C, Liu Z, Song M et al. Stabilization of heterochromatin by CLOCK promotes stem cell rejuvenation and cartilage regeneration. Cell Res 2021;31:187-205.

[105] Jing Y, Lu H, Li J et al. Vitamin C conveys geroprotection on primate ovaries. Cell Stem Cell 2025;32:1723-40.e9.

[106] Zeng X, Shi CE, Han Y et al. A metabolic atlas of blood cells in young and aged mice identifies uridine as a metabolite to rejuvenate aged hematopoietic stem cells. Nat Aging 2024;4:1477-92.

[107] Lu Y, Brommer B, Tian X et al. Reprogramming to recover youthful epigenetic information and restore vision. Nature 2020;588:124-9.

[108] Hickson LJ, Langhi Prata LGP, Bobart SA et al. Senolytics decrease senescent cells in humans: preliminary report from a clinical trial of Dasatinib plus Quercetin in individuals with diabetic kidney disease. EBioMedicine 2019;47:446-56.

[109] Lei J, Xin Z, Liu N et al. Senescence-resistant human mesenchymal progenitor cells counter aging in primates. Cell 2025;188:5039-61.e5.

[110] Ruiz JG, Oliva AA, Ramdas KN et al. Randomized phase 2b dose-escalation trial of stem cell therapy with laromestrocel for aging frailty. Cell Stem Cell 2026;33:393-404.e4.

[111] Lu YR, Tian X, Sinclair DA. The information theory of aging. Nat Aging 2023;3:1486-99.

RIGHTS & PERMISSIONS

The Author(s) 2026. Published by Oxford University Press on behalf of Higher Education Press.