Science cannot predict the exact date or pathogen behind the next pandemic, but decades of research have narrowed the field of candidates and identified the conditions that make a spillover event more likely. Respiratory viruses, particularly influenza and coronaviruses, sit at the top of most threat assessments, and the ecological, climatic, and behavioral trends that feed pandemic risk are all moving in the wrong direction. The tools for detecting and responding to threats are improving fast, but whether they can outpace the growing risk depends on investment, political coordination, and a willingness to act on warnings before a crisis arrives.
The Viral Families Scientists Watch Most Closely
Not all pathogens are equally capable of causing a pandemic. The ones that worry researchers the most share a specific combination of traits: they spread through the respiratory route, they have animal reservoirs that bring them into contact with people, and they lack reliable vaccines or treatments ready to deploy at scale. A 2025 analysis in Open Forum Infectious Diseases identified six viral families that meet these criteria: Orthomyxoviridae (which includes influenza), Coronaviridae (SARS, MERS, and their relatives), Paramyxoviridae (Nipah, Hendra), Picornaviridae, Pneumoviridae, and Adenoviridae. Each of these families already includes viruses that circulate seasonally in humans, which means the biological machinery for human-to-human spread exists within the family tree.1PubMed Central. Viral Families With Pandemic Potential
Coronaviruses, for their part, have shown a pattern of evolving rapidly once they enter a new host. Research on four different coronaviruses found that the spike gene, which governs how the virus attaches to host cells, is consistently under strong evolutionary pressure in intermediate hosts. That means coronaviruses are particularly good at fine-tuning themselves to bind to new receptors, which is exactly the kind of adaptation that enables a jump from animals to people.2Journal of Evolutionary Biology. Variable routes to genomic and host adaptation among coronaviruses
Why H5N1 Gets Special Attention
Influenza pandemics have occurred roughly three to four times per century, and the current top concern is H5N1 avian influenza. What changed the calculus was the virus jumping into dairy cattle in the United States in 2024, creating a new mammalian reservoir that brings it into closer contact with farmworkers and, by extension, the broader human population.
The gap between the current H5N1 virus and one capable of efficient human-to-human spread may be disturbingly narrow. A 2025 study in Science analyzed the hemagglutinin protein from the first reported human-infecting bovine H5N1 strain and found that it still preferred avian-type receptors. But a single amino acid substitution switched its binding preference to human-type receptors, and a second mutation enhanced that shift further.3PubMed Central. A single mutation in bovine influenza H5N1 hemagglutinin switches specificity to human receptors A broader review of H5N1 mutations catalogued several changes in viral proteins beyond hemagglutinin that could boost the virus’s ability to replicate in mammalian cells, evade human immune defenses, and resist antiviral drugs.4PubMed Central. Evolution of H5N1 Cross-Species Transmission: Adaptive Mutations Driving Avian-to-Human Infection
That said, laboratory experiments suggest the full transition to airborne transmission in mammals requires multiple coordinated mutations, not just one. Research using ferrets, the standard animal model for human flu transmission, found that even after engineering several receptor-switching mutations into H5N1, the virus could spread by direct contact but not through airborne respiratory droplets without additional reassortment with other influenza genes.5PubMed Central. In vitro evolution of H5N1 avian influenza virus toward human-type receptor specificity The complexity of changes required is somewhat reassuring, but every new mammalian host the virus circulates in gives it more chances to accumulate those mutations. And population immunity to novel influenza strains is thin. A serological study found that existing human antibody levels against swine-origin H1N2 and Eurasian avian-like H1N1 viruses would only reduce their transmission potential by roughly a fifth, meaning these strains would need a very low intrinsic transmissibility to be blocked by herd immunity alone.6PubMed Central. Determining Existing Human Population Immunity as Part of Assessing Influenza Pandemic Risk
Ecological Pressure Points
Viruses do not spontaneously decide to infect humans. Spillover events happen at interfaces where people, livestock, and wildlife overlap under conditions of stress and close contact. Three ecological trends are widening those interfaces.
Deforestation and agricultural expansion are the most studied. A systematic review found that land-use changes like deforestation, urbanization, and intensified farming accelerate the spread of zoonotic pathogens and the risk of transmission, with these pressures expected to grow as human populations expand and demand for resources increases.7PubMed Central. Emerging zoonotic diseases originating in mammals: a systematic review of effects of anthropogenic land-use change Clearing forests pushes wildlife into closer contact with people and livestock while also restructuring animal communities in ways that favor generalist species, which tend to harbor more pathogens.8PubMed Central. Land reversion and zoonotic spillover risk
Live animal markets compound the problem. These markets bring together diverse species that would rarely encounter each other in the wild, cramming them into dense, stressful environments that suppress their immune systems while creating ideal conditions for pathogens to jump between hosts and, eventually, into people passing through.9AJPM Focus. Mitigating Wildlife Spillover in the Clinical Setting: How Physicians and Veterinarians Can Help Prevent Future Disease Outbreaks
Biodiversity loss is a subtler but equally important driver. In intact ecosystems, the sheer variety of species dilutes the density of any one pathogen’s preferred hosts. Meta-analyses have confirmed that these “dilution effects” are common in natural disease systems.10PubMed Central. Dilution effects in disease ecology When biodiversity declines, species that happen to be good reservoirs for zoonotic diseases become disproportionately abundant, while the nonreservoir species that buffer against transmission disappear.11PubMed Central. Impacts of biodiversity and biodiversity loss on zoonotic diseases
Climate Change Is Redrawing the Map of Risk
Warming temperatures are not just an environmental concern; they are actively reshaping where disease-carrying animals live. Vampire bats, which transmit rabies, have expanded their range northward in the Americas over the past century in a pattern linked to changing climate conditions.12PubMed Central. Climate change linked to vampire bat expansion and rabies virus spillover And because bat species richness is strongly correlated with the number of coronaviruses circulating in a region, shifts in bat distributions carry broader implications. A study found that southern China’s Yunnan province and neighboring areas in Myanmar and Laos have become a global hotspot for climate-driven increases in bat diversity, precisely the region where SARS-like coronaviruses are most abundant.13PubMed Central. Shifts in global bat diversity suggest a possible role of climate change in the emergence of SARS-CoV-1 and SARS-CoV-2
What makes this concerning for pandemic prediction is that climate-driven range expansion is not a future scenario. It is already happening, and it is bringing pathogen-carrying wildlife into contact with human populations that have never been exposed to those pathogens before.
How Global Travel Turns Local Outbreaks Into Pandemics
A novel virus emerging in a remote village in 1920 might have burned out locally. Today it can reach any major city within hours. Research tracking the arrival times of COVID-19 and its variants in countries around the world found that a country’s air travel intensity was a significant predictor of how quickly it was reached by a new variant.14Transport Policy. The effects of air transport mobility and global connectivity on viral transmission: Lessons learned from Covid-19 and its variants Air travel can connect essentially any two points on the planet within a day or two, which means a respiratory pathogen with even a modest incubation period can be seeded across continents before anyone realizes there is an outbreak at the source.15Trends in Parasitology. Human Mobility and Global Health Security
This is not a solvable problem in the traditional sense. Nobody is going to shut down global air travel as a preventive measure. The practical implication is that early detection and rapid response must be fast enough to compensate for the speed of modern transportation networks.
Hunting for Threats Before They Emerge
The traditional model of disease surveillance, waiting for sick people to show up at hospitals, is too slow for pandemic prevention. Several newer approaches aim to catch dangerous pathogens while they are still circulating in animal populations or environmental reservoirs.
Wastewater surveillance proved its value during COVID-19 and is now being adapted for broader pathogen monitoring. A systematic review found a consistent link between the presence of pathogens in wastewater and clinical cases in surrounding communities, with regular sampling in vulnerable areas providing early warning of outbreaks.16PubMed Central. Wastewater Surveillance for Early Warning of Infectious Disease Outbreaks: A Systematic Review of Evidence and Implications for One Health Surveillance The advantage is that wastewater captures signal from an entire population, including people who never seek medical care.
On the wildlife side, researchers are cataloguing the viruses carried by animals that live near human communities. A metagenomic study of insectivores in Yunnan Province, China, profiled the viromes of shrews, hedgehogs, and moles across 12 counties and identified 68 viruses, including 57 previously unknown species.17PubMed Central. Host taxonomy and environment shapes insectivore viromes and viral spillover risks in Southwestern China A parallel effort using nanopore sequencing on urban bats in São Paulo state detected viral sequences from 12 families of public health relevance, including coronaviruses and paramyxoviruses, and even identified a previously unknown filovirus in bats in the Americas.18One Health. Viral metagenomics of synanthropic urban bats: A surveillance strategy for uncovering potentially zoonotic viruses
These studies are not just academic cataloguing exercises. The goal is to build a library of known threats so that when a new human infection appears, researchers can quickly determine whether it matches something already observed in wildlife and understand its potential behavior.
Machine Learning as an Early Warning System
One of the more promising developments in pandemic prediction is the use of machine learning models to flag viruses that are genetically primed to infect humans, before they actually do so. A convolutional neural network trained on viral protein sequences achieved roughly 97% accuracy in identifying viruses with the genetic patterns associated with human infection, tested on data the model had never seen before.19PubMed Central. Zoon0PredV: Potential Virus Species Crossover Prediction Using Convolutional Neural Networks and Viral Protein Sequence Patterns Another approach used large language models pre-trained on genomic data and then fine-tuned for infectivity prediction, outperforming existing models across most viral families tested.20Communications Medicine. Hidden challenges in evaluating spillover risk of zoonotic viruses using machine learning models
For avian influenza specifically, researchers have developed alignment-free computational methods that analyze the genetic composition of avian influenza virus proteins to classify strains as zoonotic or not, even while the virus is still circulating in birds. The idea is to identify which strains in poultry or wild birds are genetically closest to known human-infecting variants, giving health authorities a head start.21bioRxiv. Predicting the Risk of Avian Influenza Zoonosis using Viral Genome Sequencing Data
The researchers behind these tools are candid about their limitations. High accuracy on training data does not guarantee real-world predictive power when a truly novel virus appears, and the models can struggle with viral families that are underrepresented in existing databases. Still, combining these tools with field surveillance creates a layered detection system that did not exist a decade ago.
Faster Vaccines and Field-Ready Diagnostics
Even with better detection, a pandemic pathogen will eventually emerge. The speed of the response then becomes everything. The “100 Days Mission,” championed by the Coalition for Epidemic Preparedness Innovations (CEPI) and supported by the G7, sets a target of having safe and effective vaccines manufactured and ready within 100 days of identifying a pandemic pathogen. Achieving that goal depends on maintaining “warm” manufacturing capacity that can be rapidly pivoted, along with pre-negotiated regulatory pathways, secure supply chains, and improved global logistics.22PubMed Central. The 100 Days Mission: a perspective on accelerating vaccine manufacturing for future pandemics
COVID-19 exposed how fragile vaccine supply chains can be under pressure. Manufacturing capacity was a severe bottleneck, and wealthy countries absorbed the initial supply while low-income countries waited months for meaningful quantities.23PubMed Central. Resilience in the Vaccine Supply Chain: Learning from the COVID-19 Pandemic The inequity was not just a moral failure; it was a strategic one, because every country where the virus circulated unchecked was a country where new variants could emerge.
On the diagnostic side, CRISPR-based platforms are being developed as rapid, field-deployable alternatives to traditional laboratory testing. Platforms like SHERLOCK and DETECTR can detect vanishingly small amounts of pathogen genetic material with specificity comparable to PCR but without the need for a laboratory.24PubMed Central. CRISPR-Cas systems for diagnosing infectious diseases In field testing, SHERLOCK detected Zika and dengue viruses directly from patient bodily fluids in under two hours with no instruments required, and new assays for previously uncharacterized pathogens could be designed and validated within a week.25PubMed Central. Field-deployable viral diagnostics using CRISPR-Cas13 That kind of turnaround could be transformative during the early days of an outbreak, when laboratories are overwhelmed and the pathogen has not yet been fully characterized.
Antimicrobial Resistance as a Parallel Threat
Most pandemic predictions focus on viruses, but the growing crisis of antimicrobial resistance represents a different kind of pandemic risk that is already underway. Resistant bacteria do not grab headlines the way a novel virus does, but projections estimate millions of deaths annually by 2050 if current trends hold.26Journal of Pure and Applied Microbiology. The Silent Pandemic: A One Health Review of Alternative Strategies to Combat Antimicrobial Resistance And the two threats are not independent. COVID-19 accelerated antibiotic misuse in many settings, as patients received antibiotics for a viral infection and overwhelmed hospitals struggled to maintain stewardship programs.27PubMed Central. A parallel and silent emerging pandemic: Antimicrobial resistance (AMR) amid COVID-19 pandemic
The practical worry is that a future viral pandemic could be made far worse if secondary bacterial infections are untreatable. During the 1918 influenza pandemic, most deaths were actually caused by bacterial pneumonia following the initial viral infection. If the next influenza pandemic arrives in a world where key antibiotics no longer work against common respiratory bacteria, the death toll could be amplified well beyond what the virus alone would cause.
Synthetic Biology and Engineered Risks
The scenarios above involve natural spillover, but advances in synthetic biology have introduced a separate category of risk. It is now technically possible to synthesize entire viruses from scratch, including dangerous ones. Researchers have demonstrated the synthesis of poliovirus from its published genetic sequence, and in 2018 a team described methods for chemically synthesizing horsepox virus, a close relative of smallpox, for roughly $100,000. Since most of the world’s population has no immunity to smallpox, the implications were alarming.28Journal of Biosafety and Biosecurity. Challenges and recent progress in the governance of biosecurity risks in the era of synthetic biology
The dual-use dilemma is real. The same capabilities that let researchers rapidly produce SARS-CoV-2 virions for vaccine development in early 2020 could, in different hands, be used to create modified pathogens. Advances in both synthetic biology and artificial intelligence are making these capabilities increasingly accessible to people without deep specialist training.29University of Oxford. Panoptic dual-use management: preventing deliberate pandemics in an age of synthetic biology and artificial intelligence Governance frameworks have not kept pace. Screening protocols for DNA synthesis orders exist but are not universally enforced, and the rapid improvement in benchtop DNA synthesis equipment is outrunning the oversight infrastructure.
Why Public Trust Shapes Pandemic Outcomes
The pathogen determines the ceiling of a pandemic’s damage, but human behavior determines how close we get to that ceiling. Non-pharmaceutical interventions like masking, distancing, and isolation are extremely difficult to enforce at scale, making voluntary compliance critical. Research using European survey data found that both interpersonal trust and trust in institutions played a significant role in whether people followed government-mandated restrictions during COVID-19.30Administration & Society. To Obey or not to Obey? The Role of Trust in Non-pharmaceutical Interventions A separate study found that higher trust in government was associated with greater compliance with precautionary measures, and that this effect worked partly through increasing people’s awareness of the problem itself.31PubMed Central. When and How Trust in Government Leads to Compliance with COVID-19 Precautionary Measures
Communication quality matters as much as the interventions themselves. A review of COVID-era evidence on behavioral interventions concluded that even interventions shown to be effective in clinical settings become ineffective in the real world if they are not communicated clearly, meaning if people do not understand whether, when, or how to use them.32PubMed Central. Effectiveness of communications in enhancing adherence to public health behavioural interventions: a COVID-19 evidence review This is not a soft or secondary issue. The erosion of public trust in health institutions in many countries since 2020 is, from a pandemic preparedness standpoint, as dangerous as any new mutation in a virus.
Integrating Across Disciplines With One Health
Most of the drivers discussed above, deforestation, livestock practices, wildlife surveillance, antimicrobial resistance, climate-driven range shifts, fall under the jurisdictions of different agencies and ministries that rarely coordinate with one another. The One Health framework is an attempt to change that by treating human, animal, and environmental health as a single interconnected system rather than three separate domains.33PubMed Central. One Health: A Holistic Approach to Tackling Global Health Issues
In practice, One Health calls for smart surveillance at the interfaces where wildlife, livestock, and humans overlap, combined with research to speed vaccine and therapeutic development and strategies to reduce the upstream drivers of spillover risk.34PubMed Central. Pandemic origins and a One Health approach to preparedness and prevention: Solutions based on SARS-CoV-2 and other RNA viruses The concept has gained high-level political endorsement, including through ASEAN declarations and the One Health Joint Plan of Action. But an assessment of implementation in Southeast Asia, one of the highest-risk regions for zoonotic emergence, found that environmental and wildlife sectors remain underrepresented in governance and resource allocation, with most effort still focused on reactive outbreak response rather than proactive prevention.35PubMed Central. Scaling up the integration of the environment and wildlife sectors for enhanced One Health outcomes in Southeast Asia
The gap between endorsing One Health at a summit and actually funding coordinated wildlife-livestock-human surveillance networks at the local level is where pandemic preparedness tends to stall. The science for identifying threats is more advanced than it has ever been. The bottleneck is organizational and political, getting different sectors to share data, share budgets, and act on warnings generated outside their traditional lanes.