Biological surveillance does not primarily suffer from a shortage of data. It suffers from an inability to make data generated across different biological domains function as a coherent system. That is no longer simply an informatics problem. It is a biosecurity problem.
Modern biological surveillance generates information at a scale that would have been inconceivable a generation ago. Human diagnostic laboratories, veterinary surveillance programmes, genomic sequencing platforms, environmental monitoring systems, food-chain surveillance and research laboratories continuously produce observations capable of identifying emerging biological threats. Yet these systems remain fragmented by incompatible data models, inconsistent terminology, incomplete provenance and institutional boundaries. The result is a paradox: the biological world is increasingly observable while the infrastructure required to assemble those observations into coherent evidence remains incomplete.
This article argues that interoperability should therefore be understood as biosecurity infrastructure rather than simply an information-management objective. It distinguishes technical connectivity from semantic and organisational interoperability; examines why provenance, identifiers and shared standards matter during biological surveillance; considers the particular interoperability problem created by the One Health interface between human, animal and environmental systems; and argues that standards-based, distributed architectures offer a more realistic model than universal data centralisation.
The technical problem, however, has a boundary. Standards can determine how biological information is represented and exchanged. They cannot determine who has authority to share it, who is responsible for acting upon it, or how competing institutional obligations are resolved. At that point, interoperability becomes a governance problem.
Keywords
1. The Surveillance Data Paradox
Biological surveillance has undergone a remarkable technical transformation.
Sequencing that once required specialist infrastructure and substantial time can now be performed routinely. Diagnostic laboratories generate increasingly structured datasets. Veterinary surveillance systems collect observations across livestock populations. Environmental surveillance can identify biological signals in wastewater and other environmental samples. Computational systems can compare thousands or millions of observations in ways that would be impossible through manual analysis.
The intuitive conclusion is that biological surveillance should consequently be becoming progressively more integrated.
It is not clear that it is.
The problem is not necessarily the quantity of information available. It is whether observations generated in different systems can be combined while retaining enough meaning to constitute reliable evidence.
Consider an emerging infectious event.
A veterinary laboratory identifies an unusual cluster of disease. A sequencing facility characterises an organism associated with several cases. Environmental surveillance identifies related genetic material. A human diagnostic laboratory subsequently detects an organism with a similar sequence.
Biologically, these observations may describe different manifestations of the same event.
Computationally and institutionally, however, they may exist in entirely different worlds.
The veterinary records may identify animals through one classification system and locations through another. The genomic data may contain extensive sequence information but limited epidemiological metadata. Environmental records may be organised around sampling sites rather than hosts. Clinical records may be protected within healthcare infrastructure and represented using terminology developed for human medicine.
Even something apparently straightforward such as a date may describe sample collection in one system, accession in another and analysis in a third.
The data exists.
The connections between it do not necessarily exist.
This distinction has significant consequences. Surveillance is not simply the accumulation of observations. Its value comes from the ability to identify relationships between observations sufficiently early and reliably to inform action.
A surveillance infrastructure that generates excellent individual datasets but cannot combine them is not necessarily an integrated surveillance system.
It may instead be a collection of sophisticated silos.
2. Connectivity Is Not Interoperability
2.1 The File Transfer Fallacy
Interoperability is often treated as a technical integration problem.
System A contains information. System B requires it. An API, data export or integration pipeline is created between them. The systems can now exchange information and are therefore described as interoperable.
This is connectivity.
It is necessary, but it is not sufficient.
For information to be genuinely interoperable, the receiving system must be capable of interpreting what the transmitted information represents.
Suppose a record contains:
Species: Pig
Sample: Lung
Result: Positive
Date: 12 August 2026
The record is readable to a human observer. It is also profoundly underspecified.
What taxonomic identifier defines “pig”?
Does “lung” mean an entire organ, a particular anatomical location, a tissue specimen or material derived from that specimen?
Positive for what?
Using which assay?
Against what threshold?
Was 12 August the date of collection, accession, analysis, validation or reporting?
Was the animal clinically affected?
Was the result subsequently amended?
The information can be transferred perfectly while its meaning remains ambiguous.
This is why interoperability has several layers.
Technical interoperability allows systems to communicate.
Syntactic interoperability allows them to exchange information using compatible structures.
Semantic interoperability allows the receiving system to interpret the information consistently.
And organisational interoperability allows institutions to incorporate that information into their operational processes.
Biological surveillance requires all four.
2.2 Why Biology Makes the Problem Harder
Biology presents an unusually difficult interoperability problem because biological information is contextual.
The significance of an observation frequently depends upon the specimen, host, method, experimental conditions, temporal relationship and provenance surrounding it.
A nucleotide sequence without metadata is still a nucleotide sequence. Its surveillance value, however, may be dramatically reduced if its host, sampling location, collection date or analytical history cannot be established.
WHO’s guidance on pathogen genome data sharing recognises the importance of making pathogen genomic data available for preventing, detecting and responding to epidemics and pandemics and emphasises the need for data-sharing systems that are effective, equitable and trustworthy.[4]
Likewise, the statement that an organism was detected can mean very different things depending upon whether detection occurred through culture, PCR, sequencing, microscopy, serology or another analytical method.
Biological meaning therefore does not reside exclusively in the result.
It resides in the relationship between the result and its context.
An interoperable biological infrastructure has to preserve those relationships.
3. Standards, Semantics and Provenance
3.1 Standards as Infrastructure
Standards are sometimes perceived as administrative constraints imposed upon scientific work.
A more useful interpretation is that standards provide a shared language through which independently developed systems can exchange meaning.
The principle is visible throughout modern infrastructure.
TCP/IP allows networks designed and operated by different organisations to communicate. HTML allows independently developed browsers and servers to interpret web content. DICOM provides a standardised framework for exchanging medical imaging information, while HL7 FHIR provides a standard for exchanging healthcare data.[5,6]
None of these standards requires the organisations using them to become identical.
They establish an interoperable layer between them.
Biological information requires something analogous.
Different disciplines will continue to require different applications, workflows and specialised data structures. A veterinary pathology system should not become a human electronic health record. A genomic repository should not become an environmental monitoring platform.
The objective is not uniformity.
It is sufficient common structure that information crossing the boundary retains its meaning.
3.2 Ontologies and Controlled Meaning
Terminology is one part of this infrastructure.
Natural language is exceptionally effective for communication between humans because humans routinely resolve ambiguity using context.
Computational systems cannot safely make the same assumptions.
If one dataset describes Sus scrofa, another “pig”, another “porcine” and another an internal species code, a human may recognise that the records potentially refer to the same species.
A machine requires a mechanism for establishing that equivalence.
Ontologies, controlled vocabularies and persistent identifiers provide that mechanism.
The same principle applies to diseases, anatomical structures, laboratory procedures, environmental locations, biological materials and experimental processes.
The OBO Foundry illustrates the principle particularly clearly. Its purpose is not simply to create biological vocabularies but to coordinate ontologies according to shared principles so that independently developed biological datasets can be integrated.[7]
Semantic standards allow systems to move from exchanging strings to exchanging concepts.
That distinction becomes increasingly important as biological surveillance becomes computational.
3.3 Provenance Is Part of the Observation
Provenance is frequently treated as metadata surrounding a scientific result.
For surveillance purposes, it is more useful to regard provenance as part of the result itself.
A biological observation should ideally retain information concerning:
- where the specimen originated;
- when and how it was collected;
- which organism or environment it relates to;
- how it was transported or processed;
- which analytical method was used;
- which instrument or software generated the result;
- who or what validated the result;
- whether the record has subsequently changed; and
- what evidence supports its interpretation.
These questions are fundamental to reproducibility.
They are also fundamental to biosecurity.
During an emerging event, decision-makers may need to combine evidence generated by institutions operating under different quality systems and analytical protocols.
Two apparently identical results may not carry identical evidential weight.
Without provenance, that distinction can disappear.
Interoperability that preserves the result while discarding its history therefore creates only partial interoperability.
The FAIR principles provide a useful conceptual foundation. Making data Findable, Accessible, Interoperable and Reusable is not simply about making files available. The principles explicitly emphasise machine actionability and the ability of computational systems to find, access, interoperate with and reuse data with minimal human intervention.[1]
4. The One Health Interoperability Gap
The limitations of current biological information infrastructure become particularly visible when surveillance crosses the human-animal-environment interface.
One Health starts from a biological observation that is difficult to dispute: human health, animal health and environmental systems are connected.
The information systems built to observe them have developed largely independently.
Important integration initiatives now exist, but cross-domain interoperability remains incomplete. ECDC itself identifies improved interoperability and coordination across sectors, mandates and borders as an objective of its One Health framework and notes that practical cross-sector implementation remains at an early stage in many settings.[8]
Human healthcare has developed clinical information architectures around patients, healthcare encounters, diagnostic processes and medical confidentiality.
Veterinary systems have developed around animals, herds, holdings, veterinary practices, production systems and animal-disease regulation.
Environmental surveillance may be structured around geographical sites, ecological systems, water bodies, species populations or sampling programmes.
Food-chain systems introduce further infrastructures concerned with establishments, batches, movements and traceability.
Genomic surveillance creates another information layer organised around sequences, variants and phylogenetic relationships.
Each architecture is rational within its own domain.
The interoperability problem emerges at their boundaries.
A zoonotic event does not care that the veterinary record and the human-health record use different identifiers.
Antimicrobial resistance does not remain within the database in which it was detected.
A pathogen entering a food-production system does not become a fundamentally different biological object because responsibility moves from one regulatory organisation to another.
The underlying biology crosses institutional boundaries considerably more easily than the information describing it.
This creates what might be called the One Health interoperability gap: the distance between the biological connectedness of the system being observed and the informational connectedness of the institutions observing it.
Closing that gap is not simply desirable for research.
It is important to biological security.
The Quadripartite One Health Joint Plan of Action makes essentially the same institutional observation from the governance side: effective One Health requires collaboration, communication, capacity building and coordination across the sectors responsible for human, animal, plant and environmental health.[2]
5. When Data Fragmentation Becomes Biosecurity Risk
5.1 Detection Depends on Relationships
Many emerging biological threats are not immediately visible as major events.
They become visible because apparently minor observations form a pattern.
An unusual diagnosis in one location may mean little. Similar observations across several locations may indicate an emerging event.
A genetic sequence may be unremarkable in isolation. Its relationship to sequences from another host population may change its significance.
An environmental signal may be difficult to interpret until combined with clinical or veterinary observations.
Surveillance therefore depends upon relationships.
Fragmentation makes those relationships harder to identify.
The security consequence is straightforward: if relevant observations cannot be combined, the point at which an emerging pattern becomes detectable may be delayed.
In biological events characterised by rapid transmission or geographical spread, delay is not a neutral variable.
It changes the options available to those responsible for responding.
WHO’s global genomic surveillance strategy consequently places genomic surveillance within a broader architecture of epidemic and pandemic preparedness rather than treating sequencing as an isolated laboratory capability.[3]
5.2 Europe Is Already Building One Health Data Infrastructure
The interoperability problem is not hypothetical.
European institutions have begun constructing cross-sector infrastructure specifically because foodborne and zoonotic surveillance requires information generated in different domains to be analysed together.
The collaboration between the European Food Safety Authority and the European Centre for Disease Prevention and Control around whole-genome sequencing provides a useful example.
Foodborne disease illustrates precisely why conventional institutional divisions can become problematic. Human infections may be detected through healthcare systems while the relevant organism is simultaneously present in food, animals, production facilities or other parts of the food chain.
Understanding the event requires observations from both sides of that administrative divide.
The EFSA–ECDC approach to One Health whole-genome sequencing reflects recognition that genomic information becomes substantially more useful when isolates and their associated epidemiological information can be compared across human and non-human sources.
The sequence provides evidence of biological relationship.
The associated sampling and epidemiological information provides context.
The interoperable infrastructure allows the two to be considered together.
This is One Health expressed not as a slogan, but as information architecture.
It also demonstrates an important point: cross-sector interoperability does not arise automatically simply because sequencing technology improves.
It has to be deliberately built.
A second example now exists within animal-health reporting itself. The European Union’s Animal Disease Information System and WOAH’s WAHIS have been interconnected to reduce duplicate reporting and create a more streamlined exchange of outbreak information between European and global animal-health systems.[9]
These developments demonstrate both the problem and the direction of travel.
Interoperability is increasingly being treated as operational infrastructure.
5.3 The Cost of Translation
Fragmented standards also create a less dramatic but persistent cost: translation.
Whenever information moves between incompatible systems, somebody or something has to interpret it.
Sometimes that process is automated.
Frequently it remains surprisingly manual.
Spreadsheets are reformatted. Column names are changed. Identifiers are mapped. Terminology is reconciled. Metadata is reconstructed from accompanying documents. Analysts make decisions about whether two fields represent the same concept.
Each translation introduces cost.
It also introduces opportunities for error.
In routine research these inefficiencies are frustrating.
During an emerging biological event, they consume something more important: time.
5.4 Automation Amplifies Both Quality and Error
The increasing use of machine learning and automated biological analysis makes the quality of interoperability more consequential, not less.
Computational systems can identify patterns across datasets at scales impossible for human analysts.
But computational scale does not resolve semantic ambiguity.
It amplifies whatever structure is supplied to it.
If equivalent concepts are represented inconsistently, automated analysis may treat them as different.
If different concepts are represented identically, it may combine observations that should remain separate.
If provenance is absent, a model may treat observations generated under substantially different analytical conditions as equivalent.
The promise of AI-enabled biological surveillance therefore depends upon an infrastructure problem that receives considerably less attention than the models themselves.
Before biological information can become reliably machine-intelligible at scale, it must first become sufficiently structured.
6. Interoperability Does Not Require a Global Biological Database
A common response to fragmentation is centralisation.
If distributed information systems create interoperability problems, the intuitive solution is to move information into a single system.
There are circumstances where central repositories are extremely valuable.
Universal centralisation, however, is neither technically necessary nor institutionally realistic.
Biological information may be subject to clinical confidentiality, commercial sensitivity, national-security considerations, research restrictions, intellectual-property rights and jurisdiction-specific regulation.
Institutions also have legitimate reasons for retaining operational control over their information.
Countries have even stronger reasons.
The choice is therefore not between fragmentation and universal centralisation.
Standards enable a third model: federated interoperability.
Information can remain under the control of the organisation responsible for it while being represented sufficiently consistently to support defined forms of exchange, query or analysis.
A veterinary surveillance organisation does not need to surrender its entire database to a human public-health authority for the two organisations to exchange relevant observations.
An environmental surveillance platform does not need to become part of a clinical record system to communicate a biological signal.
A genomic repository does not need to own the underlying epidemiological record to maintain a persistent relationship to it.
The infrastructure requirement is the ability to preserve identity, meaning, relationships and provenance across these boundaries.
This is precisely what open standards are capable of supporting.
It also explains why interoperability and data sovereignty should not automatically be treated as opposing objectives.
Well-designed interoperability can permit institutions to retain control while still participating in a larger surveillance network.
WHO’s pathogen genome data-sharing principles illustrate the importance of this balance. Effective sharing depends not only upon making data available but upon trust, transparency, appropriate governance and consistency with applicable legal and ethical frameworks.[4]
7. From Technical Interoperability to Governance
There is, however, a limit to what standards can solve.
A standard can define how an organism is identified.
It cannot determine whether an institution is legally permitted to disclose the observation.
A standard can define how provenance is represented.
It cannot determine whether another institution trusts the organisation that generated the result.
An API can transmit an alert within milliseconds.
It cannot determine who has responsibility for acting upon it.
An ontology can establish that two systems are describing the same disease.
It cannot determine which government agency has authority when that disease crosses from wildlife into livestock and subsequently into humans.
At this point the interoperability problem changes character.
It becomes a governance problem.
Questions of authority, accountability, incentives, jurisdiction, confidentiality, reporting obligations and institutional trust sit above the technical infrastructure.
This distinction is important because there is a danger in assuming that better data integration automatically produces better biological security.
It does not.
A perfectly interoperable surveillance infrastructure could still fail if institutions are unwilling to share information, uncertain about their responsibilities or unable to coordinate action.
Conversely, excellent governance cannot extract information from systems whose data cannot be interpreted across institutional boundaries.
The two layers are interdependent.
Molecular Precision’s concern is principally with the first: the standards, semantics, provenance and computational structures that allow biological information to remain meaningful as it moves between systems.
The second belongs to the wider field of One Health governance.
Work at One Health Security examines this institutional layer: how governance, biosecurity, law, surveillance structures and institutional trust determine whether societies can act effectively upon biological information.
The distinction provides a useful conceptual boundary.
Interoperability determines whether institutions can understand one another’s biological information. Governance determines whether they can do anything with it.
A functioning biological-security architecture requires both.
8. What Biological Surveillance Infrastructure Should Look Like
If interoperability is treated as infrastructure rather than an integration project, several design principles follow.
8.1 Persistent Biological Identity
Objects should be capable of retaining identity across systems.
Specimens, organisms, sequences, experiments and derived observations require identifiers that can survive movement between applications and institutions.
Without persistent identity, relationships between records become probabilistic rather than explicit.
8.2 Explicit Semantics
Systems should describe important biological concepts using machine-readable semantics rather than relying exclusively upon local field names and human interpretation.
This does not require one universal ontology.
It does require mechanisms through which equivalence and relationships can be established.
8.3 Provenance by Design
Provenance should not be reconstructed after information becomes important.
Systems should capture it during creation.
The analytical history of an observation is part of its scientific meaning and should be represented accordingly.
8.4 Open Exchange Standards
Where mature standards exist, organisations should prefer them to proprietary representations unless there is a compelling reason not to.
A proprietary format may optimise an individual application.
An open standard optimises the ecosystem in which applications have to coexist.
These are different objectives.
Synthetic Biology Open Language provides one example from biological design. SBOL was created to provide a machine-readable representation through which biological designs can be exchanged between software tools and organisations.[10]
The precise standard required for surveillance will differ by domain. The underlying principle does not.
8.5 Federated Rather Than Necessarily Centralised Architecture
Interoperability should not depend upon universal data ownership.
Distributed systems should be capable of exchanging defined information while retaining institutional control over source datasets.
This is likely to be particularly important for international biological surveillance.
8.6 Machine-Readable by Default
The future consumer of a biological record may not be a human.
It may be a surveillance algorithm, automated laboratory, regulatory system or computational model.
Information architectures designed today should assume machine interpretation will become increasingly important.
That requires explicit structure.
9. The Strategic Opportunity
The interoperability problem is also a strategic opportunity for the biological sciences.
The organisations that define standards do more than solve technical problems.
They shape markets.
DICOM changed medical imaging not because it produced better scanners, but because it established common infrastructure through which imaging equipment, archives and software systems could exchange information.
FHIR is attempting something analogous across healthcare data exchange.
The economic value is created above the standards layer.
Standards make that value easier to connect.
Biological standards have the same infrastructural characteristic.
Open representations for biological design, genomic information, experimental provenance and surveillance observations do not eliminate proprietary innovation.
They define the layer on top of which proprietary innovation becomes easier to integrate, validate and scale.
This is why pharmaceutical companies, biotechnology organisations, diagnostic laboratories, research institutions and governments should care about biological interoperability even where their own internal systems function adequately.
The relevant question is not:
Does our system work?
It is:
Can our system participate meaningfully in the biological information environment that is emerging around it?
Those are very different standards of success.
The European Commission’s wider data strategy reflects a similar economic logic: greater value can be generated when data can move within interoperable environments while appropriate rights and controls remain in place.[11]
Biology will increasingly face the same choice.
Organisations can continue constructing isolated applications and subsequently pay the cost of connecting them.
Or interoperability can become a design assumption.
10. Conclusion
Biological surveillance does not lack information.
It lacks sufficient infrastructure for making independently generated information function collectively.
The distinction matters.
Human health, animal health and environmental systems increasingly observe different parts of the same biological world. Sequencing allows relationships between organisms to be identified with unprecedented precision. Environmental surveillance creates new ways of detecting biological signals. Computational methods make it possible to analyse relationships across datasets at enormous scale.
Yet these capabilities remain constrained by an older infrastructure problem.
Systems use different terminology.
Identifiers do not persist across boundaries.
Provenance is incomplete.
Equivalent observations are represented differently.
Institutional systems cannot reliably interpret one another’s records.
The result is an interoperability gap between the biological connectedness of the world and the informational connectedness of the systems attempting to observe it.
That gap has consequences for reproducibility and research efficiency.
It also has consequences for biosecurity.
Standards cannot solve the entire problem. They cannot create institutional trust, determine jurisdiction or establish who has responsibility for acting on an emerging threat.
But without standards, those institutions may never receive information in a form they can reliably interpret in the first place.
Biological security therefore has at least two infrastructures.
One is institutional: governance, authority, law, surveillance responsibility and trust.
The other is informational: identifiers, standards, semantics, provenance and interoperability.
The two cannot sensibly be separated.
As biological surveillance becomes more distributed, more computational and more international, an information architecture assembled piecemeal around individual institutional requirements will become increasingly difficult to sustain.
Interoperability is not an administrative convenience.
It is not merely an API problem.
And it is not simply a desirable property of scientific data.
It is part of the infrastructure through which modern biosecurity functions.
References
- Wilkinson, M.D., Dumontier, M., Aalbersberg, I.J. et al. (2016). The FAIR Guiding Principles for scientific data management and stewardship. Scientific Data, 3, 160018.
- World Health Organization, Food and Agriculture Organization of the United Nations, United Nations Environment Programme & World Organisation for Animal Health (2022). One Health Joint Plan of Action (2022–2026): Working Together for the Health of Humans, Animals, Plants and the Environment. Geneva: World Health Organization. ISBN 978-92-4-005913-9.
- World Health Organization (2022). Global Genomic Surveillance Strategy for Pathogens with Pandemic and Epidemic Potential, 2022–2032. Geneva: World Health Organization.
- World Health Organization (2022). WHO Guiding Principles for Pathogen Genome Data Sharing. Geneva: World Health Organization. ISBN 978-92-4-006174-3.
- Health Level Seven International (2023). FHIR Release 5 (Version 5.0.0). HL7 International.
- DICOM Standards Committee. Digital Imaging and Communications in Medicine (DICOM) Standard. Medical Imaging & Technology Alliance/National Electrical Manufacturers Association.
- Smith, B., Ashburner, M., Rosse, C. et al. (2007). The OBO Foundry: coordinated evolution of ontologies to support biomedical data integration. Nature Biotechnology, 25, 1251–1255.
- European Centre for Disease Prevention and Control (2024). ECDC One Health Framework. Stockholm: ECDC.
- World Organisation for Animal Health (2023). ADIS, the EU Animal Disease Information System connected to WAHIS. Paris: WOAH.
- Baig, H., Fontanarrosa, P., Kulkarni, V. et al. (2020). Synthetic Biology Open Language (SBOL) Version 3.0.0. Journal of Integrative Bioinformatics, 17(2–3), 20200017.
- European Commission (2020). A European Strategy for Data. COM(2020) 66 final. Brussels: European Commission.
- World Organisation for Animal Health. World Animal Health Information System (WAHIS). Paris: WOAH.