Medical Entomology as a One Health Information Science: Connecting Vector Ecology, Pathogen Detection, Host Vulnerability, and Environmental Surveillance

INTRODUCTION

Medical entomology increasingly operates within a surveillance environment in which changes in vector distribution, pathogen movement, climate suitability, urban growth, trade, travel, land use, and host contact can interact across multiple scales. Arbovirus emergence, for example, is rarely attributable to a single environmental or biological cause; it more often reflects the convergence of vector competence, population abundance, susceptible hosts, ecological opportunity, pathogen introduction, and mobility-mediated exposure. Information systems that isolate one of these dimensions may accurately document a component of risk while failing to represent the process through which that component becomes epidemiologically consequential. A One Health information-science perspective is therefore needed not simply to combine records, but to preserve the biological meaning, spatial context, temporal relevance, and inferential limits of each record as it moves from observation to interpretation [1].

The surveillance problem also changes over the course of vector-borne disease emergence. Evidence that is informative during an initial phase of possible introduction may be insufficient once a vector has become locally established or when sporadic pathogen detections begin to occur. Adaptive surveillance frameworks developed for emerging tick-borne zoonoses illustrate why objectives, sampling strategies, geographical resolution, and response expectations should evolve with the stage of emergence rather than remain fixed [2]. This principle is applicable beyond ticks: detecting an arthropod in a new locality, documenting repeated seasonal occurrence, establishing reproductive persistence, identifying pathogen infection, and confirming human or animal transmission represent distinct evidentiary states. An information architecture must therefore encode these states separately so that surveillance outputs do not collapse occurrence, establishment, infection, infectiousness, exposure, and disease into a single undifferentiated category of “risk.”

The decision value of medical-entomology surveillance further depends on coordination among entomological, veterinary, environmental, laboratory, epidemiological, and public-health functions. Cross-sector collaboration is not an optional administrative addition to surveillance; it determines whether observations collected by one sector can be interpreted and acted upon by another. Work linking vector-borne disease surveillance with broader sustainable-development and One Health objectives shows that surveillance systems must connect ecological monitoring to defined decision needs, institutional responsibilities, and intervention capacity [3]. This requirement shifts the analytical focus from whether data are available to whether the data are biologically interpretable, semantically aligned, timely for the intended decision, and accompanied by enough contextual information to support or limit inference.

New modes of distributed observation may expand surveillance reach but also demonstrate why data acquisition alone cannot solve the architecture problem. Citizen-science mosquito surveillance in a resource-constrained setting showed that community participation can increase geographical coverage and local engagement, yet the usefulness of the resulting observations still depends on specimen identification, participant retention, reporting consistency, spatial representation, and integration with professional surveillance [4]. Comparable limitations arise with automated traps, remote sensing, genomic sequencing, syndromic indicators, and digital mobility data. The objective of this article is consequently to develop a proposed One Health medical-entomology information architecture that connects vector ecology, pathogen detection, host vulnerability, environmental observation, and decision processes while preserving the distinctions among empirical observation, biological interpretation, transmission inference, and operational action.

Fragmentation in contemporary medical-entomology surveillance

Fragmentation first appears institutionally. Entomological units may monitor vector presence and abundance, diagnostic laboratories may detect pathogens, environmental agencies may collect climatic or land-cover data, and human or animal health services may document cases without a shared structure for defining questions, exchanging metadata, or interpreting discordant signals. A matrix developed to evaluate multisectoral One Health surveillance demonstrates that collaboration must be assessed across organizational, functional, and operational dimensions rather than inferred from the nominal participation of multiple sectors [5]. Participatory surveillance-design research likewise indicates that stakeholders need to agree on system objectives, expected outputs, responsibilities, information flows, and decision uses before technical integration can become meaningful [6]. Without that agreement, datasets may be merged while the underlying surveillance purposes remain incompatible.

Fragmentation also occurs conceptually because sectors classify related observations differently. A mosquito record may be treated as an ecological occurrence, an invasive-species alert, an operational control trigger, or an indirect disease-risk indicator. A positive pathogen assay may be viewed as laboratory confirmation of infection in a captured arthropod, although public-health interpretation may silently elevate it to evidence of local transmission. One Health surveillance-development guidance emphasizes that effective systems require coordinated detection, verification, assessment, communication, and response across human, animal, and environmental domains [7]. Systems leadership and governance scholarship similarly shows that implementation depends on reciprocal relationships among institutional capacity, shared understanding, leadership, and public engagement [8]. These findings support an architecture in which each observation retains its original construct definition and provenance while its permissible interpretations are specified separately.

A third form of fragmentation is evaluative. Surveillance programmes frequently report outputs such as numbers of traps, specimens, tests, reports, meetings, or participating agencies, but these measures do not necessarily demonstrate better detection, earlier warning, more accurate risk classification, or more appropriate intervention. A systematic review of evaluation tools applicable to One Health surveillance found that existing approaches remain heterogeneous and often require adaptation to the complexity of integrated systems [9]. The absence of common evaluation logic makes it difficult to compare systems or determine whether apparent integration improves decisions. A defensible architecture must therefore link each surveillance component to a stated function, expected information product, uncertainty condition, and validation requirement rather than assuming that greater connectivity is inherently beneficial.

The evidence dimensions and interpretive boundaries for fragmentation in contemporary medical-entomology surveillance are summarized in Table 1.

Table 1. Fragmentation in Contemporary Medical-Entomology Surveillance: Vector Systems, Biological Mechanisms, Exposure Pathways, Evidence Requirements, Uncertainty, and Interpretive Boundaries

Vector ecology and population intelligence

Vector population intelligence requires more than a static map of reported occurrence. It should characterize whether a vector population is present, reproductively established, seasonally active, increasing, dispersing, or changing in ways that alter contact with hosts. Local environmental variables can influence these processes differently. Modelling of Aedes aegypti abundance has shown that relative humidity and precipitation-driven egg hatching can improve representation of population dynamics, but the resulting relationship remains dependent on local climate, container ecology, sampling design, and the biological lag between rainfall and adult emergence [10]. Temperature similarly affects multiple components of Zika virus transmission, including mosquito survival, development, biting, and viral incubation, producing nonlinear effects that cannot be represented by a simple assumption that warmer conditions uniformly increase risk [11]. Population intelligence must therefore preserve the mechanisms and time lags connecting environmental exposure to observed vector outcomes.

Thermal suitability is especially useful for illustrating the distinction between mechanistic knowledge and realized transmission. Comparative work on mosquito-borne disease demonstrates that transmission-related traits often operate within bounded and pathogen-specific temperature ranges [12]. Climate-driven projections of changing Aedes-borne virus suitability can identify areas where environmental conditions may become more permissive, but these projections do not establish that competent vectors, pathogens, susceptible hosts, and effective contact will coexist at the required place and time [13]. Environmental suitability should consequently be represented as one contextual layer within the architecture, not as a substitute for entomological observation or epidemiological confirmation. The same principle applies to land cover, drought, vegetation, water storage, urban form, and other environmental variables: each may alter vector ecology, yet its interpretation depends on species-specific biology and the local pathways through which people or animals encounter vectors.

Operational forecasting can extend population intelligence from retrospective description to prospective situational awareness. Forecasts of relative Aedes abundance have been developed to assist mosquito-control operations by combining surveillance and environmental information [14]. Such approaches are potentially valuable because they can identify periods or locations where intensified observation may be warranted. Nevertheless, forecast output remains conditional on model specification, calibration data, surveillance continuity, transferability, and the decision horizon for which it was designed. A population forecast is not itself a transmission forecast, and neither becomes an operational rule until performance has been assessed prospectively under the intended conditions of use. The proposed architecture therefore treats population models as qualified analytical products accompanied by provenance, validation history, uncertainty, and a clearly stated relationship to the decisions they are intended to inform.

Pathogen detection and transmission indicators

Pathogen surveillance in arthropods can identify circulation earlier than clinical case detection, reveal unexpected microbial diversity, and direct attention to vector–pathogen combinations that warrant further investigation. Advances in mosquito-borne arbovirus surveillance include targeted molecular assays, virus isolation, high-throughput sequencing, saliva-based approaches, and other strategies for detecting rare signals within large collections [15]. Metabarcoding can simultaneously support mosquito identification and broad biological characterization, increasing efficiency where morphological identification is difficult or specimen volumes are high [16]. These methods expand what surveillance can observe, but their information value depends on whether the detected material represents contamination, a blood-meal remnant, an infected vector tissue, a disseminated infection, or a pathogen capable of being transmitted.

Metatranscriptomic surveillance further illustrates the trade-off between breadth and inferential specificity. Its sensitivity and specificity can support broad detection of known and unexpected viruses, yet performance is affected by RNA quality, sequencing depth, bioinformatic classification, background organisms, and confirmatory testing [17]. Xenosurveillance, in which blood-feeding arthropods are analysed as samplers of vertebrate-associated pathogens, can reflect some signals identified through conventional human sampling and may be useful where access to clinical surveillance is constrained [18]. However, a pathogen sequence recovered from a mosquito blood meal provides information about pathogen presence in a vertebrate host population, not necessarily about infection of the mosquito or its capacity to transmit that pathogen. The architecture must therefore record specimen type, tissue or pool composition, feeding status, analytical method, confidence of identification, and the biological state that the assay can legitimately support.

Transmission indicators should be constructed as an evidentiary progression rather than a binary extension of pathogen detection. Relevant states may include pathogen presence in the environment or host population, acquisition by a vector, infection of the midgut, dissemination beyond the gut, salivary-gland infection, infectious saliva, compatible vector–host contact, and observed human or animal cases. Different methods resolve different positions along this progression. A positive pooled assay may justify intensified sampling or confirmatory investigation, but it cannot alone determine where infection was acquired, whether the vector was infectious, or whether local transmission occurred. Likewise, absence of detection may reflect low prevalence, inadequate sample size, unsuitable timing, assay limitations, or incomplete spatial coverage rather than true absence. A One Health information architecture should consequently retain pathogen detection as a high-value but bounded observation, attach explicit uncertainty and alternative explanations, and require additional entomological, host, temporal, and epidemiological evidence before elevating the signal to a claim of transmission.

Host vulnerability and exposure context

Vector presence and environmental suitability become epidemiologically meaningful only when they intersect with susceptible hosts under conditions that permit contact. Human mobility can connect areas of vector abundance with populations that would otherwise remain spatially separated, altering where exposure occurs and where infections are subsequently detected. Analyses of urban mosquito-borne disease systems show that commuting, travel, and routine movement can redistribute contact beyond residential neighbourhoods and complicate attribution of infection to a single location [19]. An information architecture should therefore avoid treating residence as an automatic proxy for exposure. It should distinguish place of residence, place of work or study, travel history, activity space, vector habitat, probable infection window, and diagnostic location whenever the supporting data permit that resolution.

Socioeconomic and environmental conditions further shape exposure through housing quality, water storage, waste management, vegetation, occupational activity, access to prevention, and health-service use. Comparative dengue research in Chinese cities indicates that natural and socioeconomic determinants may contribute differently across urban contexts, meaning that a factor associated with transmission in one city may not retain the same importance elsewhere [20]. Bayesian spatial analysis in tropical Australia similarly demonstrates that climate variability interacts with socio-ecological conditions rather than operating independently of them [21]. These findings argue against a universal vulnerability index assembled from context-free indicators. Host vulnerability should instead be represented as a conditional construct whose components, measurement scales, causal assumptions, and applicability are recorded explicitly.

Human infectiousness is another frequently missing layer. Vector infection risk depends not only on the existence of reported cases but also on when infected people are viraemic, how infectiousness varies across the course of illness, whether competent vectors feed during that interval, and whether subsequent vector survival permits transmission. Direct mosquito feeding on people infected with dengue virus has demonstrated that human-to-mosquito infectiousness is dynamic and cannot be inferred solely from a binary clinical diagnosis [22]. Clinical case counts therefore remain essential but incomplete transmission indicators. A One Health information architecture should connect case data to symptom onset, sampling date, likely exposure interval, pathogen characteristics, host infectiousness evidence, vector activity, and spatial uncertainty while avoiding the unsupported inference that every reported infection generated local onward transmission.

Environmental observation and early warning

Environmental observation can improve surveillance when variables are connected to biologically plausible mechanisms and interpreted over suitable time lags. Rainfall may create larval habitat in one setting while flushing breeding sites in another; drought may suppress some aquatic habitats yet increase domestic water storage; temperature may accelerate development within a viable range but reduce survival beyond physiological limits. Modelling of dengue in Barbados identified nonlinear and delayed climatic associations, illustrating that contemporaneous correlations may miss the timing through which environmental conditions affect vectors, pathogens, and host contact [23]. Environmental data should therefore enter the architecture with explicit spatial resolution, temporal aggregation, lag structure, source, preprocessing history, and the mechanism through which each variable is expected to influence a surveillance outcome.

Forecasting systems can combine environmental and epidemiological information to estimate near-term disease risk, but predictive success is not automatically transferable across seasons, locations, or decision contexts. A probabilistic seasonal dengue-forecasting system developed for Vietnam used a superensemble approach to accommodate multiple model structures and uncertainty sources [24]. Dynamic ensemble methods applied in Brazil likewise incorporated weather and population-susceptibility cycles to forecast epidemic years [25]. These studies support the value of model diversity and probabilistic output, but they also show why a forecast must be interpreted relative to its training period, prediction horizon, calibration, validation geography, surveillance quality, and intended operational response. A model that discriminates high- and low-risk periods may still be unsuitable for determining when or where a particular control action should begin.

Early warning becomes operationally meaningful only when environmental and entomological signals are connected to a defined decision process. A nationwide Brazilian case study used primary-health-care data to support earlier detection of dengue outbreaks, demonstrating the potential value of routinely collected health information when transformed into timely surveillance indicators [26]. Nevertheless, an early-warning signal is not equivalent to a validated decision rule. The usefulness of an alert depends on timeliness, false-alarm costs, missed-event consequences, action capacity, communication pathways, and prospective evidence that acting on the signal improves an identified outcome. The architecture should therefore distinguish raw observations, derived indicators, model-generated alerts, analyst interpretation, authorized escalation, and completed action, with performance evaluated at each transition.

The evidence dimensions and interpretive boundaries for environmental observation and early warning are summarized in Table 2.

Table 2. Environmental Observation and Early Warning: Vector Systems, Biological Mechanisms, Exposure Pathways, Evidence Requirements, Uncertainty, and Interpretive Boundaries

Proposed one health information architecture

The proposed architecture begins with an observation layer that keeps environmental, entomological, pathogen, host, and health-system records distinct while attaching provenance and context to each item. A contemporary One Health data-integration framework for real-time pathogen surveillance and applied genomic epidemiology demonstrates the value of connecting heterogeneous sources through defined workflows, shared infrastructure, and cross-sector interpretation [27]. For medical entomology, the observation layer should record what was measured, where, when, by whom, with which protocol, and under which sampling conditions. It should also record what was not observed only when sampling effort and detection limits make non-detection interpretable. Environmental suitability, trap capture, species confirmation, breeding evidence, pathogen assay results, host exposure, clinical diagnosis, and genomic linkage should remain separate evidence objects rather than being prematurely compressed into a single risk score.

The second layer provides biological and semantic interpretation. Technical connectivity allows data to move, but interoperability requires that receiving systems preserve the meaning of variables, units, classifications, and relationships. Digital-health scholarship has emphasized that interoperability includes technical, syntactic, semantic, and organizational dimensions rather than mere data exchange [28]. In the proposed architecture, each evidence object is therefore linked to an ontology or controlled definition, a spatial and temporal scale, an inferential category, and an uncertainty statement. A vector-presence record may support occurrence; repeated records plus immature stages may support establishment; a pathogen-positive vector may support infection or circulation; salivary evidence may strengthen infectiousness inference; and locally acquired compatible cases may support transmission. These states are connected, but no state is automatically substituted for the next.

Figure 1 depicts the integrated information architecture within the analytical logic developed in this section.

Alt text

A structured conceptual diagram that depicts the integrated information architecture, with labelled components, directional relations, contextual modifiers, uncertainty points, and a clear boundary between observed evidence and proposed synthesis.

The third layer converts classified evidence into qualified decision signals. FAIR maturity frameworks show why data should be findable, accessible, interoperable, and reusable through transparent, community-governed assessment rather than through unexamined claims of openness [29]. Community approaches to standards and repositories further demonstrate that reusable information depends on shared descriptions of data structures, reporting expectations, and domain conventions [30]. Within this architecture, a decision signal is generated only after the relevant evidence has passed defined checks for provenance, comparability, biological meaning, spatial–temporal compatibility, uncertainty, and decision relevance. The resulting signal should identify its evidentiary basis and specify whether it supports routine observation, intensified surveillance, confirmatory testing, targeted communication, intervention consideration, or formal incident assessment.

Figure 2 shows how environmental, entomological, pathogen, and host data become decision signals within the analytical logic developed in this section.

Alt text

A structured conceptual diagram that shows how environmental, entomological, pathogen, and host data become decision signals, with labelled components, directional relations, contextual modifiers, uncertainty points, and a clear boundary between observed evidence and proposed synthesis.

The fourth layer is governance and learning. A generalizable One Health framework for zoonotic-disease control highlights the need to connect surveillance with coordinated assessment, prevention, response, and evaluation across sectors [31]. The architecture therefore includes named decision owners, escalation criteria, documentation of dissent or uncertainty, version control for definitions and models, feedback from interventions, and periodic reassessment of whether the information products remain useful. This layer prevents analytical outputs from being detached from accountability. It also creates a learning cycle in which false alarms, missed detections, data gaps, implementation constraints, and unexpected outcomes become inputs for redesign rather than being hidden within a final risk category.

The proposed components, evidence bases, boundary conditions, failure modes, and validation requirements are organized in Table 3.

Table 3. Proposed One Health Information Architecture: Components, Evidence Basis, Relations, Boundary Conditions, Failure Modes, and Validation Requirements