AI in Healthcare: 10 Powerful ways it can fight the next pandemic
Explore how AI in healthcare can detect outbreaks early, track virus spread, and accelerate treatments
Spoiler alert: COVID-19 was just a warm-up. Michael Osterholm, renowned epidemiologist and director of the Center for Infectious Disease Research and Policy, has just published “The Big One”, a chilling book that paradoxically opens the door to the most concrete technological hope we have: artificial intelligence applied to public health.
While politicians are still congratulating themselves on their “exemplary management” of COVID-19, Osterholm delivers a disturbing truth: we failed miserably against a virus that, on the pandemic scale, was just a gentle kitten compared to the tiger waiting for us. Osterholm and his co-author Mark Olshaker explore how a future pandemic caused by a virus even deadlier than SARS-CoV-2 could unfold. But where others see apocalypse, I see an unprecedented opportunity for AI to revolutionize our approach to global health crises.
The Big One: How We Must Prepare for Future Deadly Pandemics
A must-read for anyone who cares about the future of our planet.
The Programmed Failure of Our Current Systems: A Brutal Reality Check
Osterholm argues in his new book that COVID-19 was just a dress rehearsal for what could happen. This claim, far from being alarmist, rests on stubborn facts. COVID-19 had a relatively low mortality rate (less than 1%) and primarily affected older people. Imagine for a moment a pathogen with Ebola’s lethality (up to 90%) and SARS-CoV-2’s transmission. The result? Civilizational collapse, nothing more, nothing less.
Our current early detection systems resemble blind sentries posted at the borders of a declining empire. The WHO, supposed to be our global watchdog, took weeks to acknowledge human-to-human transmission of COVID-19. Meanwhile, a Canadian startup, BlueDot, had detected the warning signals nine days before the UN organization. This temporal difference counts in thousands of lives. In the context of a truly devastating pandemic, these few days could make the difference between survival and extinction.
Osterholm’s book highlights a fascinating paradox: we today possess technologies of unprecedented power, capable of predicting our shopping preferences or recognizing our face in a crowd, but we remain unable to effectively anticipate and counter pandemic threats. This technological schizophrenia is not inevitable. Rather, it reveals our collective inability to orient innovation toward the issues that truly matter.
AI as Planetary Nervous System: Early Detection Reimagined
Artificial intelligence can revolutionize our approach to epidemiological surveillance in a revolutionary way. Where our current systems analyze fragmented data with weeks of delay, AI can process massive and heterogeneous information flows in real-time: hospital data, social networks, Google searches, mobility data, environmental signals, and even satellite imagery.
Take the concrete example of HealthMap, developed by Boston Children’s Hospital. This AI system automatically analyzes millions of multilingual information sources: official reports, local media, discussion forums, healthcare professional testimonials. By cross-referencing this data with natural language processing algorithms and machine learning, HealthMap can identify the emergence of local epidemics days or even weeks before traditional surveillance systems.
But the most promising innovation lies in predictive analysis of viral mutations. MIT researchers have developed AI models capable of predicting viral genetic evolution with stunning accuracy. These systems analyze viral genomic sequences in real-time and identify potentially dangerous mutations before they even spread massively. Imagine being able to anticipate the emergence of a vaccine-escape variant not in reaction, but in anticipation.
Federated learning represents another silent revolution. This technology allows AI algorithms to learn from distributed data without centralizing sensitive information. Concretely, a hospital in Wuhan, another in Milan, and a third in New York can make their respective AIs collaborate to detect emerging pandemic patterns, while preserving the confidentiality of their patient data. This approach circumvents the geopolitical and regulatory obstacles that paralyzed international cooperation during COVID-19.
Pandemic Modeling: When AI in Healthcare Becomes Oracle
Osterholm emphasizes that preparation will be crucial, particularly in the field of vaccines. Current epidemiological models, based on classical differential equations, resemble road maps compared to modern GPS systems. AI can integrate thousands of variables in real-time: human behaviors, government measures, environmental factors, economic data, transport networks, and even public sentiment analyzed via social networks.
The GLEAM model (Global Epidemic and Mobility), developed by the ISI Foundation, uses machine learning techniques to simulate pathogen spread on a planetary scale. This system can model the impact of specific containment measures in particular cities on global pandemic dynamics. Even more impressive, it can virtually test thousands of intervention strategies and identify optimal combinations of health, economic, and social measures.
These AI models can also simulate complex counterfactual scenarios. What would have happened if China had shared its genomic data a week earlier? If Europe had closed its borders ten days before? If vaccines had been distributed according to epidemiological rather than geopolitical logic? These questions, impossible to treat with classical models, become accessible thanks to AI. This retrospective simulation capability allows for continuous refinement of our future strategies.
AI can also model the impact of social determinants of health on pandemic spread. An algorithm developed by Stanford University analyzes how socio-economic inequalities, access to care, urban density, and even cultural beliefs influence epidemic dynamics. This holistic approach goes beyond the purely biomedical vision of pandemics to integrate their profoundly social and political dimension.
Therapeutic Acceleration: AI at the Heart of the Race Against Time
As the authors emphasize, COVID-19 killed more than 7 million people worldwide, and this figure could have been dramatically reduced if we had effective treatments available more quickly. Artificial intelligence is already revolutionizing drug discovery, but its potential in pandemic contexts remains largely underexploited.
Atomwise, a California startup, uses AI to identify therapeutic molecules in days rather than years. Their algorithm virtually analyzes millions of chemical compounds and predicts their effectiveness against specific pathogenic targets. For COVID-19, Atomwise identified several promising drug candidates in less than a week after the virus structure was published. This temporal acceleration could make the difference between a controlled pandemic and global chaos.
AlphaFold, DeepMind’s AI, has revolutionized our understanding of protein structures. By predicting the three-dimensional shape of viral proteins with near-perfect accuracy, AlphaFold dramatically accelerates vaccine and treatment development. Moderna used this technology to design its COVID-19 vaccine in just 48 hours after receiving the viral genetic sequence. Imagine this speed applied to a truly lethal pathogen.
AI can also optimize clinical protocols themselves. Adaptive learning algorithms can analyze clinical trial results in real-time and automatically modify doses, inclusion criteria, or study arms to maximize effectiveness and minimize risks. This “smart trial” approach could halve the time needed to validate new treatments.
Even more revolutionary, AI can design personalized therapies on a large scale. By analyzing patients’ genomes, their medical histories, their immune response, and the specific characteristics of the pathogen, algorithms can recommend tailored treatments. This pandemic precision medicine would transform our therapeutic approach from a “one size fits all” model to an individualized and optimized strategy.
Hospital Orchestration: AI as Crisis Conductor
Hospital systems were the first victims of COVID-19, overwhelmed by an influx of patients they could neither predict nor absorb efficiently. AI can transform this chaotic reactivity into strategic anticipation. Predictive algorithms can analyze local demographic data, infection rates, comorbidities, and accurately predict needs for ICU beds, ventilators, and medical staff.
Johns Hopkins Hospital already uses an AI system called CRISP that predicts the clinical evolution of COVID-19 patients with 95% accuracy. This system analyzes vital signs, biological results, and medical histories in real-time to identify patients at risk of deterioration. This anticipation allows for early care and optimizes allocation of scarce resources.
AI can also revolutionize hospital logistics. Optimization algorithms can coordinate in real-time patient transfers between facilities, supply of medications and medical equipment, and even planning of medical teams. Amazon has developed similar systems to optimize its global supply chain; these technologies can be adapted to the specific constraints of the healthcare system in crisis situations.
More subtly, AI can optimize the architecture of care itself. By analyzing patient flows, nosocomial infection risks, and the effectiveness of different care pathways, algorithms can recommend spatial and organizational reorganizations in real-time. This dynamic adaptability transforms hospitals from rigid structures into living organisms capable of evolving according to epidemic constraints.
Fighting Misinformation: AI as Guardian of Truth
Osterholm rightly points to the toxicity of misinformation during health crises. COVID-19 demonstrated that an “infodemic” can be as dangerous as a biological pandemic. False information about miracle treatments, conspiracy theories, and politicization of health measures directly contributed to thousands of avoidable deaths.
Artificial intelligence can deploy sophisticated defenses against this informational pollution. Fake news detection algorithms, trained on massive corpora of verified data, can identify and flag misleading content in real-time. These systems analyze not only textual content, but also diffusion patterns, funding sources, and even stylistic characteristics of messages to detect coordinated misinformation campaigns.
More innovatively, AI can generate personalized and culturally adapted counter-narratives. Rather than brutally censoring false information, algorithms can create targeted awareness messages, using the same linguistic register and cultural references as the targeted audiences. This informational “jujitsu” approach uses the force of misinformation against itself.
Medical chatbots powered by AI represent another crucial line of defense. These systems can instantly respond to public health questions, provide updated and reliable information, and even identify users potentially victims of misinformation to offer them personalized support. The conversational assistant developed by the WHO during COVID-19 answered more than 10 million questions, demonstrating public appetite for reliable and accessible information sources.
Generative AI: Revolution in Crisis Communication
Health authorities failed miserably in their communication during COVID-19, alternating between contradictory messages, incomprehensible technical jargon, and public infantilization. Generative AI can revolutionize this communication by creating messages adapted to different audiences: education level, native language, cultural beliefs, and specific concerns.
An AI system can analyze in real-time the effectiveness of public health messages via social networks, opinion surveys, and observed behavioral changes. This feedback loop allows for continuous adjustment of communication strategies to maximize their impact. Imagine prevention messages that automatically evolve according to their reception by different demographic groups.
AI can also massively personalize health recommendations. Rather than generic guidelines (“respect barrier gestures”), algorithms can generate specific advice adapted to each individual’s profile: age, profession, family situation, health status, and even personality. This hyper-personalization dramatically increases adherence to preventive measures.
More revolutionary, AI can create interactive simulations allowing the public to viscerally understand the impact of their individual choices on collective pandemic dynamics. These epidemiological “serious games” transform health education from a passive exercise into an immersive and memorable experience.
Ethical and Technological Challenges: The Dark Zones of Pandemic AI
This optimistic technological vision must not obscure the considerable challenges that await us. The use of AI in pandemic contexts raises complex ethical questions about mass surveillance, personal data protection, and algorithmic bias. An AI system trained on predominantly Western data risks performing poorly in African or Asian contexts, reproducing and amplifying global health inequalities.
Technological dependence also represents a systemic risk. What happens if digital infrastructures collapse during a major crisis? How can we guarantee the resilience of hyper-complex AI systems against cyberattacks deliberately targeting our pandemic response capabilities? These health cybersecurity questions become crucial in a world where information warfare has become an extension of classical geopolitics.
Social acceptability perhaps constitutes the most underestimated challenge. COVID-19 revealed deep distrust toward scientific expertise and public health institutions. How can we convince already skeptical populations to accept recommendations generated by “black box” algorithms? Transparency and explainability of AI become major democratic issues.
Governance and International Cooperation: Rethinking the Global Health Architecture
As Osterholm and Olshaker emphasize, the next pandemic could be much worse, requiring international cooperation of unprecedented scale. AI can facilitate this coordination by creating interoperable data standards, automated information sharing protocols, and collaborative decision-support systems.
The World Health Organization should lead the creation of a “global health AI” – a distributed but coordinated system of algorithms collaborating in real-time to detect, model, and combat pandemic threats. This global digital infrastructure could transcend national rivalries by creating a technological common good serving planetary health.
Public-private partnerships become crucial for mobilizing technological innovation in service of public health. Google, Microsoft, Amazon have computing capabilities and technical expertise that most governments cannot match. How to structure these collaborations to maximize effectiveness while preserving the general interest? This question will largely define our collective capacity to face future health crises.
Training and Transformation of Health Professionals
Integrating AI into pandemic response requires a radical transformation of medical and public health training. Future epidemiologists will need to master massive data analysis, machine learning, and interpretation of complex algorithms. This pedagogical revolution is already underway in certain American universities offering hybrid medicine-computer science curricula.
More fundamentally, we must rethink medical culture itself. Traditional medicine, based on individual clinical experience and intuition, must hybridize with a data-driven and algorithmic approach. This cultural transition, more than technological, will largely determine the success of AI in public health.
Health professionals must also develop new communication skills to explain AI-generated recommendations to the public. How to popularize a complex algorithmic decision without losing scientific credibility or fueling conspiracy theories? This AI medical pedagogy becomes an essential clinical skill.
Toward Planetary Predictive Medicine
AI offers us the historic opportunity to move from reactive medicine to predictive medicine on a planetary scale. Rather than enduring pandemics as unpredictable natural disasters, we can anticipate them, model them, and deploy preventive countermeasures. This epistemological transformation – from reaction to prediction – could revolutionize our collective relationship to health crises.
This vision is not science fiction. The technologies already exist, scattered in research laboratories, innovative startups, and R&D departments of technology companies. The challenge is no longer technical but organizational and political: how to mobilize these innovations in service of a collective project of global health protection?
AI will never replace clinical wisdom, human empathy, and social solidarity that found medicine and public health. But it can augment them, accelerate them, and multiply them on an unimaginable scale just a decade ago. This technological augmentation of our collective capacities perhaps represents our best hope against the health challenges of the 21st century.
Conclusion: The Urgency to Act Before “The Big One”
Osterholm reflects on pandemic lessons in “The Big One” and his conclusion is unequivocal: we must prepare now or tomorrow endure a catastrophe of civilizational magnitude. Artificial intelligence is not a magic solution, but it represents our best tool to transform this preparation from a wishful thinking into a concrete and effective strategy.
Time is running out. Each month of delay in developing and deploying these health AI technologies brings us closer to the moment when they can no longer save us. The massive investments devoted to commercial AI – facial recognition, targeted advertising, autonomous cars – should at least be matched by similar efforts in health AI. It’s a question of civilizational priorities.
We have a choice. We can continue to see AI as a technological gadget or recognize it as the most powerful tool ever developed to protect human health on a planetary scale. We can endure “The Big One” as our ancestors endured plague epidemics, or we can anticipate it, model it, and deploy against it all the power of artificial intelligence.
The future of humanity could well depend on this choice. And we must make it now, before it’s too late.
Sources:
- “The Big One” by Michael Osterholm
- Star Tribune – Michael Osterholm Interview on the Next Pandemic
- MPR News – Osterholm Reflects on “The Big One” Lessons
Artificial intelligence healthcare, AI pandemic response, early epidemic detection, AI virus modeling, algorithmic epidemiological surveillance, AI vaccine development, hospital optimization artificial intelligence, medical misinformation AI combat, pandemic prediction technology, public health AI systems, Big One Osterholm, pandemic preparedness AI, predictive health algorithms, machine learning medicine, AI early warning systems
