A structured framework for translating biological, environmental, and agricultural variables into a composite risk signal, and for evaluating how policy interventions modify that risk.
Introduction
If biological risk is a problem of governance, then governance requires tools for reasoning about that risk in a structured and reproducible way. The companion analysis to this piece argues that the central limitation of the One Health framework is institutional rather than scientific: the difficulty lies not in recognising that human, animal, and environmental health are interdependent, but in organising decision-making to act on that interdependence.
This Applied Research project addresses that gap from a practical direction. It presents a scenario-modelling framework that takes the variables shaping zoonotic and biosecurity risk, combines them into a single interpretable risk signal, and tests how that signal responds to different policy choices. The objective is not prediction in the conventional sense, but disciplined comparison: a way of asking, transparently, what a given intervention is likely to change and why.
The framework is deliberately conceptual and analytical rather than a production-ready system. It prioritises interpretability and structured reasoning over predictive complexity, and is designed to connect with the biological data infrastructure provided by The BioChain.
What the Model Represents
Zoonotic risk does not arise from any single cause. It accumulates across a set of overlapping conditions: how densely livestock are kept, how often wild and domestic animals come into contact, the quality of local water systems, the reach of vaccination and veterinary services, the extent of human exposure, and the strength of farm-level biosecurity. Each of these is measurable in principle, but each is typically assessed in isolation, within the administrative domain responsible for it.
The model brings these variables into a common analytical frame. Seven variables are represented:
Livestock density, wildlife contact, water quality, vaccination coverage, veterinary access, human exposure, and farm biosecurity.
Each variable is normalised onto a common scale and combined into a weighted composite, producing a single regional risk score. Expressing risk as one interpretable figure does not collapse the underlying complexity; rather, it makes the relative contribution of each factor visible, and allows regions and scenarios to be compared on consistent terms.
Policy Scenarios
A risk score is of limited value on its own. Its purpose here is comparative: to show how the composite signal shifts when policy changes. The framework models five scenarios, ordered from least to most coordinated:
Baseline represents current conditions with no additional intervention. The remaining four model distinct policy directions: improved biosecurity at the farm level, vaccination expansion, increased veterinary access, and an integrated One Health intervention that combines measures across sectors rather than pursuing them in isolation.
The integrated scenario is the analytically significant one. It tests the central claim of the One Health Security argument directly: that coordinated action across health, agriculture, and environmental domains produces a different and more favourable risk profile than the sum of the same measures applied separately. Modelling each scenario against the baseline makes that proposition examinable rather than merely asserted.
Outputs
The framework produces three categories of output. Regional risk scores summarise the composite signal for each area under analysis. Comparative scenario outputs show how those scores change across the five policy scenarios, isolating the effect of each intervention. Visual representations of the risk distribution then make the pattern legible at a glance, supporting communication with audiences who are not engaged in the modelling itself.
Together these outputs are intended to support structured, reproducible decision-making under conditions of uncertainty. They do not remove the uncertainty; they organise it, so that the assumptions behind any conclusion remain visible and open to challenge.
Methodology
The model is deterministic and scenario-based. Variables are normalised and combined into a weighted composite, and scenarios are expressed as defined adjustments to those inputs. This is a deliberate design choice. A more elaborate probabilistic or machine-learning approach might offer greater apparent precision, but at the cost of interpretability, which is precisely the quality a governance tool most requires. A decision-maker must be able to see why a risk score is what it is, and what a given intervention is assumed to do.
The framework is implemented in R, with scripts run in a transparent sequence: data generation, risk scoring, scenario evaluation, and visualisation. Each stage is separable and inspectable, so that the path from input to output can be followed and reproduced. The code is published openly:
View the repository on GitHub.
Interpretation and Limitations
The framework should be read as a structured method of reasoning, not as a forecast. Its value lies in making the structure of a risk argument explicit: which variables are considered, how they are weighted, what each policy scenario assumes, and how sensitive the result is to those assumptions. Where data are weak or contested, the model surfaces that weakness rather than obscuring it.
This interpretability-first design reflects the wider purpose of the project. The challenge identified in the conceptual work is institutional adaptation: detection capacity has advanced faster than the institutional capacity to act on what is detected. A tool that produces opaque outputs does little to close that gap. A tool that makes its reasoning legible, and that can be interrogated by people working across different sectors, is a modest but genuine contribution to coordinated decision-making.
Conclusion
One Health Security reframes biological risk as a problem of governance. This framework is one practical answer to the question that reframing raises: how can risk that emerges across ecological, agricultural, and economic domains be reasoned about coherently, and how can the effect of policy be assessed before it is committed? It does so by combining seven variables into a single interpretable signal, testing that signal against five policy scenarios, and presenting the results transparently. The intention is not to predict the future, but to support better-structured decisions in the present. The full methodology and code are available to examine, adapt, and critique
Questions & Answers
Is this framework meant to predict outbreaks?
No. The objective is not prediction in the conventional sense, but disciplined comparison — a way of asking transparently what a given policy intervention is likely to change and why.
What variables does the model combine?
Seven: livestock density, wildlife contact, water quality, vaccination coverage, veterinary access, human exposure and farm biosecurity, normalised onto a common scale and combined into a weighted composite risk score.
What policy scenarios does it test?
Five, ordered from least to most coordinated: baseline, improved farm-level biosecurity, vaccination expansion, increased veterinary access, and an integrated One Health intervention combining measures across sectors.
Why use a deterministic model rather than machine learning?
Because interpretability is the quality a governance tool most requires. A more elaborate probabilistic or machine-learning approach might offer greater apparent precision, but at the cost of a decision-maker being able to see why a risk score is what it is and what an intervention is assumed to do.
Is the code available to check?
Yes. The framework is implemented in R with scripts run in a transparent, inspectable sequence, and the code is published openly on GitHub.
What does the “integrated” scenario actually test?
It tests the central One Health Security claim directly: that coordinated action across health, agriculture and environmental domains produces a different, more favourable risk profile than the same measures applied separately.