The work, led by Professor Kwang-Hyun Cho of KAIST’s Department of Bio and Brain Engineering, introduces a technology called ROOT, short for “Revelation Of the Original circuit of irreversible Transition.” The framework is designed to locate the small set of regulatory interactions that keeps a cell locked in a changed state. By identifying and manipulating these circuits, the researchers demonstrated two conceptually different ways to control cellular memory: one that returns a cell to its former condition and another that removes the underlying mechanism that makes the change permanent.

Cells constantly alter their behavior in response to signals from their environment. A developing cell may receive instructions to specialize, an immune cell may mature into a functional subtype, or an epithelial cell may acquire the invasive properties associated with cancer. These transitions are often useful and necessary. Their persistence allows cells to maintain an identity after the initiating signal disappears. However, the same type of stability can become dangerous when a cell adopts a disease-associated state, such as the migratory and invasive phenotype generated during epithelial–mesenchymal transition, or EMT.

The molecular basis of this persistence lies in regulatory networks made from genes, proteins and signaling molecules. Within these networks, positive feedback loops allow molecules to reinforce their own activity indirectly. One molecule activates another, which activates a third, eventually returning to stimulate the original molecule. Such feedback can function like a biological latch: a temporary signal pushes the system across a threshold, after which the network continues operating without further outside instruction. According to the KAIST team, intracellular networks may contain more than a thousand interconnected positive feedback loops, making it extremely difficult to determine which ones are essential for maintaining an irreversible transition.

ROOT addresses this problem by converting molecular regulatory relationships into computational logic models. In these models, the activity of one component can be represented as dependent on the activity of others, much like logical conditions in a digital circuit. The researchers then analyzed the behavior of the resulting networks using systems biology methods, simulating how a cell responds to a stimulus, how the response persists after the stimulus is removed, and which interactions are indispensable for that persistence. This process allowed them to distinguish the extensive network surrounding an irreversible transition from the smaller group of interactions that actually sustains it.

The team describes that critical group as the “irreversibility kernel.” The kernel is not necessarily the largest or most obvious part of a regulatory network. Instead, it represents the causal circuitry that keeps the system from returning to its original state. Removing or altering the right components of the kernel can therefore have a much greater effect than broadly suppressing many unrelated molecules. The approach offers a way to move beyond correlation, where researchers observe that two molecules change together, toward a more mechanistic understanding of which molecular relationships are responsible for locking a cell into a particular fate.

The researchers proposed two control strategies based on the kernel. Resetting control returns a cell to its previous state while leaving its capacity for irreversible behavior intact. In the team’s analogy, this is like opening a locked door without dismantling the lock. A cell reset in this way could potentially be returned to an earlier identity, but it might still undergo another permanent transition if the relevant stimulus were applied again. Reversing control takes a more fundamental approach by disrupting the source of irreversibility itself. Once the responsible circuitry is disabled, the cell could, in principle, move between states more freely instead of automatically stabilizing in one direction.

To test the framework, the researchers applied ROOT to several biological systems. These included B-cell differentiation, EMT in lung cancer, and models of enterocyte and beta-cell differentiation constructed from single-cell transcriptome data. Single-cell transcriptomics measures gene activity in individual cells, allowing researchers to examine the regulatory programs associated with distinct cellular states rather than averaging signals across an entire tissue. In these models, ROOT identified causal circuits that corresponded with known determinants of cell fate and suggested more efficient strategies for resetting state transitions. The results indicate that the method can be applied not only to idealized computational networks but also to models grounded in experimental biological data.

The findings could influence how scientists think about diseases in which cells become abnormally fixed in a particular identity. Current cancer treatments often aim to kill malignant cells, while many approaches to age-related dysfunction focus on removing damaged cells or slowing the processes that produced them. A technology capable of pinpointing and manipulating the circuits that maintain abnormal states could eventually support a different strategy: restoring cells to a healthier condition. That possibility remains experimental, and computationally predicted control circuits would need to be validated in living cells, tissues and organisms before any therapeutic use could be considered. Still, by exposing the architecture of cellular irreversibility, ROOT provides a new map for investigating how biological systems remember change—and how those memories might one day be rewritten.

Subject of Research: Molecular circuits governing irreversible cellular state transitions and computational strategies for restoring or reversing those states.

Article Title: The structural origin of irreversible transitions in biological networks

News Publication Date: 21-Aug-2026

Web References: https://doi.org/10.1073/pnas.2600800123

References: Proceedings of the National Academy of Sciences; DOI: 10.1073/pnas.2600800123

Image Credits: KAIST

Keywords: ROOT framework, irreversible cell-state transitions, molecular regulatory networks, irreversibility kernel, systems biology, cancer research, aging research, epithelial–mesenchymal transition, cell differentiation, single-cell transcriptomics, cellular reprogramming

Demographic shifts are reshaping global society at an unprecedented scale. According to the United Nations’ 2022 World Population Prospects, individuals aged 60 and above have eclipsed the one billion mark worldwide and are projected to reach 2.1 billion by 2050, encompassing 22 percent of the global population. This demographic revolution brings with it multifaceted challenges: escalating chronic disease incidence, surging healthcare costs, alterations in labor markets, and amplified strain on social welfare infrastructures. These exigencies underscore an urgent need for effective strategies that address the biological underpinnings of aging and mitigate age-related pathologies.

Investment trends mirror the heightened attention to aging research and therapeutic development. In 2024 alone, funding for longevity science soared to an estimated $8.5 billion, marking a staggering 220 percent increase compared to 2023. Pharmaceutical giants are notably accelerating commitments; Roche has pledged $50 billion to research and development with a pronounced focus on aging biology, while Novartis has actively pursued acquisitions in the senotherapeutics arena. This infusion of capital is catalyzing collaborations and fostering innovation, shifting longevity science from a niche interest toward a central pillar of modern healthcare.

The ARDD conference stands as a microcosm of this burgeoning ecosystem. Established in 2014, ARDD has evolved into a premier international forum, attracting pioneering investigators and industry leaders who showcase transformative technologies and novel therapeutic paradigms. The 2025 meeting’s early sell-out status attests to the escalating demand for cross-disciplinary exchange in aging biology, biogerontology, and drug discovery. Attendees will engage with the latest advancements spanning cell simulation, systems biology, and the identification of deep aging biomarkers.

Central to the conference’s scientific discourse is Insilico Medicine, a trailblazer in the application of artificial intelligence (AI) to aging and drug discovery. As an early supporter and founder sponsor of ARDD, Insilico Medicine has consistently delivered breakthrough research. The company’s AI-driven platforms leverage deep generative models, reinforcement learning, and transformer architectures to map biological aging signatures, identify dual-purpose therapeutic targets, and design novel geroprotective molecules. Their proprietary PreciousGPT series exemplifies this approach by modeling complex biological changes across developmental and senescent stages in multiple tissues and species.

One of Insilico Medicine’s hallmark contributions lies in the development of the In Silico Pathway Activation Network Decomposition Analysis (iPANDA), a sophisticated AI methodology that deconvolutes gene expression matrices to identify disease-relevant pathways and biomarkers. This innovation has been instrumental in the transition toward data-driven biomarker discovery, seamlessly integrating multi-omic data to enhance precision in target identification. Subsequent evolution of the platform into PandaOmics incorporated over 20 advanced AI models, yielding 145 promising dual-purpose aging targets with therapeutic potential—results fully detailed in peer-reviewed publications.

The translational impact of these discoveries is illustrated by Insilico’s work on Rentosertib (ISM001-055), a first-in-class investigational compound targeting the TNIK kinase, a critical regulator implicated in multiple aging hallmarks. Rentosertib recently demonstrated dose-dependent efficacy signals in a Phase IIa clinical trial for idiopathic pulmonary fibrosis (IPF), an age-associated lung pathology. This landmark achievement not only validates AI-enabled target discovery but also exemplifies the convergence of computational biology with clinical development to address refractory age-related diseases.

Scientific exploration of aging mechanisms continues to expand. Insilico’s recent publication in Aging provided in-depth AI-driven analyses elucidating the molecular interplay between senescence pathways and IPF progression. By harnessing integrative computational frameworks, these studies offer unprecedented insights into complex pathophysiological cascades and support the rational design of geroprotective interventions that may ultimately enhance tissue resilience and repair.

Beyond discovery, Insilico’s collaborations with global partners—including Taisho, Lineage, Juvenescence, and the Buck Institute—have fortified its commitment to advancing target identification and therapeutic pipelines across multiple modalities. Additionally, alliances with Life Extension and SRW have broadened efforts into novel supplement development, underscoring the company’s holistic approach to healthspan extension encompassing both pharmaceutical and nutraceutical domains.

At ARDD 2025, Insilico Medicine will play a prominent role, with founder and CEO Dr. Alex Zhavoronkov scheduled to engage alongside luminaries from Eli Lilly, Biogen, and Fosun Pharma in a panel discussion on the pharmaceutical industry’s evolving role in longevity research. Furthermore, Dr. Zhavoronkov will deliver a keynote on the rational design of geroprotectors and clinical protocols. Senior scientist Fedor Galkin will also present novel AI-discovered targets relevant to IPF and aging, highlighting the symposium’s integration of computational innovations and translational science.

This convergence of AI, biology, and medicine embodies the future trajectory of aging research—one where computational tools accelerate hypothesis generation, target selection, and drug design, ultimately reducing the time and costs of bringing effective therapies to patients. By deepening mechanistic understanding and fostering interdisciplinary collaboration, the field inches closer to the elusive goal of healthy longevity: extending not just lifespan, but healthspan, thereby enabling individuals to live longer, healthier, and more productive lives.

Insilico Medicine’s pioneering work exemplifies how leveraging next-generation AI technologies can unlock previously inaccessible realms of biological complexity. Their platforms integrate innovations such as deep generative adversarial models, reinforcement learning algorithms, and transformer architectures, offering a robust pipeline from molecular discovery to clinical candidate identification. This paradigm shift is poised to redefine the drug discovery process, particularly for complex age-associated diseases where traditional methodologies have struggled.

As the global community confronts the realities of an aging world, initiatives like ARDD and organizations like Insilico Medicine embody optimism and pragmatic ambition. Through sustained investment, scientific rigor, and technological innovation, the biomedical community is set to transform aging from an inexorable decline into a manageable and treatable condition. The 2025 ARDD meeting will not only spotlight these advances but also galvanize the next generation of researchers and entrepreneurs to realize the promise of healthy longevity for all.

Subject of Research: Aging biology, longevity science, AI-driven drug discovery, and therapeutic development for age-related diseases.

Article Title: Charting the Future of Healthy Longevity: Insights from ARDD 2025 and Insilico Medicine’s AI Revolution

News Publication Date: Not specified

Web References:

• https://agingpharma.org/
• https://www.nature.com/nature-index/research-leaders/2025/institution/healthcare/all/global
• https://insilico.com/
• https://pharma.ai/precious
• https://www.nature.com/articles/ncomms13427
• https://pharma.ai/pandaomics
• https://doi.org/10.18632/aging.203960
• https://www.prnewswire.com/news-releases/insilico-medicine-announces-nature-medicine-publication-of-phase-iia-results-evaluating-rentosertib-the-novel-tnik-inhibitor-for-idiopathic-pulmonary-fibrosis-ipf-discovered-and-designed-with-a-pioneering-ai-approach-302472070.html
• https://www.aging-us.com/article/206295

References:

• Zhavoronkov, A. et al. Identification of dual-purpose aging targets using AI platforms (Aging, 2022).
• Rentosertib Phase IIa clinical trial results published in Nature Medicine.
• iPANDA method described in Nature Communications (2016).

Image Credits: Insilico Medicine

Keywords: Health and medicine, aging research, drug discovery, longevity, artificial intelligence, geroprotectors, idiopathic pulmonary fibrosis, AI-driven biomarker discovery, translational medicine