Trend AnalysisInterdisciplinary

One Health and AI: Bridging the Pandemic Preparedness Gap

The One Health paradigm—recognizing that human, animal, and environmental health are interconnected—is gaining urgency as zoonotic spillover events accelerate. AI offers tools for integrated surveillance, but implementation in resource-limited settings faces substantial gaps.

By Sean K.S. Shin
This blog summarizes research trends based on published paper abstracts. Specific numbers or findings may contain inaccuracies. For scholarly rigor, always consult the original papers cited in each post.

COVID-19, Ebola, MERS, avian influenza, Nipah virus—the list of zoonotic diseases that have jumped from animals to humans keeps growing. The One Health paradigm, which recognizes that human health, animal health, and environmental health are deeply interconnected, has moved from an academic framework to a policy imperative. But recognizing the interconnection and acting on it are different things. Integrated surveillance—monitoring pathogens across human, animal, and environmental domains simultaneously—remains the exception rather than the rule. AI offers tools to help, but the gap between technical capability and field implementation is wide.

The Research Landscape

Viral Emergence and One Health

Sikkema and Koopmans (2025), with 7 citations and published in Nature Reviews Microbiology, provide the most authoritative overview. Their analysis maps the pathways through which viruses emerge: wildlife reservoirs → intermediate animal hosts → human populations, with environmental change (deforestation, agricultural intensification, climate shifts) accelerating each step.

The One Health insight is that surveillance at the animal-human interface—monitoring livestock markets, wildlife populations, and agricultural workers—is more effective for early detection than monitoring human clinical cases alone. By the time a novel pathogen produces enough human cases to trigger clinical surveillance, the opportunity for containment has often passed.

The paper identifies several surveillance gaps:

  • Geographic: High-risk regions (tropical forests, intensive livestock areas) are often in countries with limited surveillance infrastructure.
  • Disciplinary: Human health, veterinary, and environmental surveillance systems typically operate independently, with different data formats, reporting chains, and funding streams.
  • Temporal: Current systems detect outbreaks after they occur. Predictive surveillance—identifying conditions that favor spillover before it happens—requires integrating environmental, ecological, and genomic data.

AI for Resource-Limited Settings

Mukherjee, Sagar, and Sikkema & Koopmans (2025) examine how AI can address the integration gap specifically in resource-limited settings—where the need is greatest and the infrastructure is weakest. Their review focuses on three AI applications:

Genomic surveillance acceleration. AI tools can analyze pathogen genomic sequences in near real-time, identifying mutations of concern and tracing transmission chains. Cloud-based analysis platforms reduce the need for local computational infrastructure.

Environmental risk modeling. Machine learning models that integrate satellite imagery (deforestation, urbanization), climate data (temperature, precipitation), and ecological surveys (wildlife distribution, livestock density) can identify geographic hotspots where spillover risk is elevated.

Cross-domain data integration. Natural language processing tools can extract structured information from unstructured surveillance reports across human health, veterinary, and environmental monitoring systems—bridging the data format and language barriers that currently prevent integration.

The paper is notably honest about limitations: AI tools require training data, and data from resource-limited settings is often sparse, incomplete, or biased toward outbreaks that have already been detected. Predictive models trained on data from well-resourced settings may not generalize to different ecological and social contexts.

H5N1 in Dairy Cattle: A Case Study

Kamel and Almagharbeh (2025), with 12 citations, analyze the 2024 emergence of highly pathogenic avian influenza H5N1 in US dairy cattle—an event that challenged existing surveillance paradigms. H5N1 was previously understood as an avian pathogen with limited mammalian transmission. Its appearance in dairy herds—animals not previously considered at risk—exposed a gap in surveillance: veterinary monitoring of dairy cattle was not looking for avian influenza, and poultry surveillance was not monitoring dairy operations.

The case illustrates the practical challenge of One Health surveillance: pathogens do not respect the disciplinary boundaries between avian virology, bovine veterinary medicine, and public health. An integrated surveillance system would have flagged the cattle infections earlier by detecting the virus in environmental samples (wastewater, barn air) before clinical illness appeared.

China's Integrated Surveillance System

Li, Li, and Kamel et al. (2025), with 1 citation, describe China's progress toward integrated One Health surveillance, documenting both achievements and remaining challenges. China's system, developed partly in response to SARS (2003) and COVID-19, integrates human disease reporting (through the National Notifiable Disease Surveillance System), livestock surveillance (through the veterinary system), and environmental monitoring (through vector surveillance programs).

The achievements are real: China's system detected and responded to several zoonotic outbreaks faster than previous systems would have allowed. But the challenges are also significant: data sharing between human and veterinary systems remains incomplete, environmental monitoring is concentrated in urban areas, and the early warning models generate many false alarms—reducing trust in the system among frontline health workers.

Critical Analysis: Claims and Evidence

<
ClaimEvidenceVerdict
Surveillance at the animal-human interface enables earlier detection than clinical surveillance aloneSikkema & Koopmans' review + H5N1 case study✅ Supported
AI can accelerate genomic surveillance and environmental risk modelingMukherjee et al.'s review of AI applications✅ Supported — technically feasible; deployment in resource-limited settings lags
Pathogens cross surveillance boundaries in unexpected waysKamel et al.'s H5N1 in dairy cattle analysis✅ Supported — concrete case study
Integrated surveillance reduces outbreak response timeLi et al.'s description of China's system⚠️ Uncertain — improvements documented but system still generates high false alarm rates

Open Questions

  • Data sharing governance: Integrated surveillance requires sharing data across ministries (health, agriculture, environment). In many countries, institutional territoriality is the primary barrier—not technology.
  • Equity in AI deployment: If AI surveillance tools are developed and hosted by Northern institutions, does this create a new form of health data dependency for the Global South?
  • False alarm management: Predictive surveillance that flags many false positives erodes trust. How do we calibrate sensitivity to maintain both early detection and credibility?
  • Wildlife surveillance: Monitoring wildlife pathogens is logistically difficult—sampling wild animals is harder than sampling livestock or hospital patients. What cost-effective approaches can scale?
  • What This Means for Your Research

    For public health researchers, the H5N1 dairy cattle case is a compelling argument for broadening surveillance beyond expected hosts. The next zoonotic threat may emerge where we are not looking.

    For AI researchers, One Health surveillance offers a challenging real-world application: heterogeneous data, sparse labels, high-stakes decisions, and resource constraints.

    Explore related work through ORAA ResearchBrain.

    References (4)

    [1] Sikkema, R.S. & Koopmans, M. (2025). Viral emergence and pandemic preparedness in a One Health framework. Nature Reviews Microbiology.
    [2] Mukherjee, D., Sagar, K., & Kobialka, R. (2025). Filling the gap: artificial intelligence-driven one health integration to strengthen pandemic preparedness in resource-limited settings. Frontiers in Public Health.
    [3] Kamel, M., Aleya, S., & Almagharbeh, W. (2025). The emergence of highly pathogenic avian influenza H5N1 in dairy cattle. European Journal of Clinical Microbiology & Infectious Diseases.
    [4] Li, Z., Li, D., & Dong, J. (2025). Integrated health surveillance and early warning systems in China under the One Health perspective. Science in One Health, 100132.

    Explore this topic deeper

    Search 290M+ papers, detect research gaps, and find what hasn't been studied yet.

    Click to remove unwanted keywords

    Search 7 keywords →