One Exposome model aims to merge toxicology, AI and environmental health into a single system
The proposed One Exposome framework treats environmental health as a connected biological problem rather than a series of isolated hazards. In this view, human disease, animal illness, and ecosystem injury are not separate events competing for attention; they are linked consequences of the same exposure pressures moving across food systems, air, water, soil, workplaces, and living communities. The framework brings together One Health, exposomics, new approach methodologies (NAMs), and artificial intelligence into one conceptual structure designed to move toxicology away from late-stage damage control and toward earlier prediction.
The paper argues that current environmental health practice still leans too heavily on compartmentalized assessment. Single-substance testing, retrospective epidemiology, and siloed surveillance can miss the cumulative effects of low-dose, mixed, and chronic exposures that act over time. This matters because many environmental threats do not behave like a single drug with a single target. They interact with biology across scales, producing effects that accumulate slowly and often appear only after the damage is already established.
One sentence captures the central problem: exposure does not stay in one compartment.
Why the exposome changes the framing of environmental risk
Exposomics provides the measurement logic behind the framework. Instead of focusing only on a known pollutant or one occupational hazard, the exposome seeks to characterize cumulative exposures across the life course, including the complex and repeated contacts that shape biological responses. That shift is important because environmental risk rarely arrives as a single event. It builds through mixtures, timing, dose, and vulnerability, all of which can change the eventual health effect.
By pairing the exposome with One Health, the framework broadens the ethical and scientific lens. Human health cannot be separated from animal health or environmental integrity when antimicrobial resistance, endocrine disruption, food-chain contamination, and climate-linked stressors move across sectors. A factory, a farm, a clinic, and a watershed can all contribute to the same biological outcome. The review presents this interdependence as the rationale for integrating environmental monitoring, veterinary intelligence, and human health data rather than managing each domain on its own.
NAMs and microphysiological systems add mechanistic depth
The technical engine of the One Exposome model comes from new approach methodologies. These include human-relevant tools such as microphysiological systems, which can reproduce aspects of organ biology outside the body and help identify toxic effects earlier than conventional approaches. NAMs are presented as a way to generate mechanistic evidence that is more directly useful for adverse outcome pathway-based reasoning, where exposure is linked step by step to molecular changes, cellular dysfunction, tissue injury, and broader health consequences.
This mechanistic emphasis is not a minor methodological preference. Environmental health decisions are often slowed by uncertainty when regulators and clinicians face complex exposures with incomplete animal-to-human translation. Human-relevant experimental platforms can reduce that gap by offering evidence anchored in biology that is closer to the patient and the population at risk. The framework also supports earlier hazard identification, which is especially valuable when waiting for overt disease would mean waiting too long.
A second hard reality follows from that logic: by the time population harm is obvious, the exposure network is already entrenched.
Artificial intelligence as the integrator of mixed evidence streams
Artificial intelligence is positioned as the connective layer that can combine data from multiple sources and convert them into actionable environmental health intelligence. The review describes AI as useful for integrating exposure data, mechanistic toxicology, multi-omics signals, and surveillance outputs across species and settings. In practice, that means AI can help identify patterns that would be difficult to detect with conventional analysis, especially when exposures are numerous, nonlinear, and biologically intertwined.
The paper places particular weight on interoperable data systems. Without common standards and linked databases, exposure findings remain trapped in separate silos: environmental monitoring here, veterinary surveillance there, clinical observations elsewhere. AI can only function well if the underlying data can be compared, harmonized, and interpreted across domains. The framework therefore assumes not just better algorithms, but better infrastructure for data sharing and translation.
This has direct public health consequences. Sentinel surveillance and integrated monitoring become more powerful when they can detect shared risks across environmental, food, veterinary, and human health sectors before those risks harden into chronic disease burdens or ecological damage.
Prevention depends on governance, not only technology
The review does not treat the One Exposome as a laboratory concept alone. It also calls for governance reform so that agricultural practice, environmental management, and clinical stewardship can operate inside a single preventive logic. Antimicrobial resistance is a clear example. Resistant organisms emerge through biological selection pressures that involve medicine, livestock management, environmental release, and human transmission. A fragmented response leaves each sector partially informed and fully vulnerable.
Endocrine disruption presents a similar challenge. The effects of endocrine-active chemicals may depend on timing, dose, developmental stage, and co-exposures. Traditional risk models often struggle with those variables because they were built for simpler toxicological questions. The One Exposome framework instead accepts complexity as the starting point and then uses systems biology, mechanistic testing, and computational integration to make that complexity measurable.
Clinical public health policy would change meaningfully if this model became operational. Exposure prevention would move upstream, and risk assessment would no longer wait for disease to appear in hospitals, clinics, or registries. The practical consequence is a more anticipatory model of health protection, one that treats environmental signals as early clinical warnings rather than background noise.
The main limitation is not concept, but execution
The framework is ambitious, but its success depends on whether surveillance systems, laboratory platforms, and governance structures can actually be linked across sectors. Interoperability is easy to recommend and difficult to build. Data quality, regulatory alignment, and cross-domain trust will determine whether the One Exposome becomes a working model or remains a useful abstraction.
Even so, the paper makes a persuasive case that environmental public health can no longer rely on reactive, single-chemical thinking when biological harm is produced by mixtures, timing, and cumulative exposures. The strongest contribution of the framework is its insistence that prevention requires more than one discipline. It requires one shared exposure architecture, one interpretive system for biology, and one coordinated strategy for action across human, animal, and environmental health.