Welcome back to Healthy Innovations! 👋
Over the past few weeks, I have been following the hantavirus outbreak linked to the MV Hondius cruise ship. At least three people have died, and WHO has coordinated monitoring and repatriation of passengers and crew across multiple countries. Officials say the strain involved is the Andes virus, which only rarely spreads between humans and typically requires close contact. It is a reminder that outbreaks rarely arrive when we are fully prepared.
So what if we could spot outbreaks earlier? A new generation of surveillance platforms is trying to do exactly that. This issue looks at what they can detect, how they work, and where the limits still are.
Let’s dive in!
The old way of tracking outbreaks
When I studied Infectious Diseases at university, the system for tracking emerging threats looked roughly like this: a clinician noticed something unusual, sent samples to a lab, waited for results, filed a report, and that report worked its way up through a chain of health authorities.
If it reached WHO in a timely fashion, an alert might follow. The whole process could take days to weeks, depending on where in the world the outbreak was emerging.
In the early days of any outbreak, weeks are exactly what you don't have.
The news cycle right now reads like a roll-call of infectious disease threats – hantavirus killing passengers aboard a cruise ship in the Atlantic, Ebola resurgent in the Democratic Republic of Congo, Mpox being detected in China, and COVID still shape-shifting across the globe. Each pathogen has its own biology, its own narrow window for containment.
What they share is a dependence on surveillance systems designed for a slower world.
A new generation of AI-driven platforms is changing that – scanning millions of data sources in real time, flagging signals that no human team could catch at speed or scale.
Beating WHO to the punch
The most cited proof of concept happened on December 30, 2019.
Just after midnight, BlueDot – a Toronto-based platform founded by infectious disease physician Dr. Kamran Khan – picked up Chinese-language news reports describing a cluster of unusual pneumonia cases near a market in Wuhan. The system flagged it, epidemiologists reviewed it, and clients received an alert the following day.
BlueDot alerted clients days before WHO's first public communications in early January 2020.
How does BlueDot work?
The platform scans news across 40+ languages alongside airline ticketing data, animal disease reports, climate data, and public health bulletins. In the Wuhan case, it didn't just detect the cluster – it predicted the cities most likely to receive early case imports: Bangkok, Hong Kong, Tokyo, Taipei, Seoul, and Singapore.
Many of the cities it ranked highest were among the first outside China to report cases.
That was unlikely to be coincidence. It was pattern recognition at a speed and scale no human team could match. Today, BlueDot monitors 190+ diseases continuously, with human epidemiologists reviewing every flagged signal before it reaches clients in government, aviation, and healthcare.
Image source: BlueDot
EPIWATCH, developed at the Kirby Institute at UNSW Sydney by Professor Raina MacIntyre, takes a similar open-source approach – scanning millions of social media posts, local news feeds, and public health reports across 46 languages for early epidemic signals.
In 2023, EPIWATCH reported early signals of the first-ever Marburg virus outbreak in Equatorial Guinea ahead of WHO's official declaration. Published research in Global Biosecurity (2025) reported evidence suggesting the platform was generating meaningful alerts in advance of traditional surveillance channels.
Professor Raina MacIntyre. Image source: EPIWATCH
What makes these platforms powerful?
No single data source. The signal emerges from combining sources most global health agencies don't monitor together:
Social media chatter in local languages
Agricultural animal disease reports
Flight data mapping how a pathogen could move
Local news from conflict zones and low- and middle-income countries (LMICs)
Critically, EPIWATCH is open-source – built to make this early warning capability available in precisely the settings where traditional surveillance infrastructure collapses.
The gap that technology alone won't close
None of this progress dissolves the structural problem at the centre of global outbreak risk.
The pathogens most likely to trigger the next pandemic will emerge in LMICs – precisely the places where surveillance infrastructure is weakest.
A 2025 review in Microorganisms framing epidemic intelligence through the lens of Disease X put it plainly: surveillance limitations in LMICs are not a local problem. They are a global security vulnerability.
Disease X is a placeholder term for a hypothetical future pathogen that could cause a serious epidemic or pandemic. It does not refer to a known disease; the idea is used by the WHO and others to help governments and scientists prepare for the next unknown threat.
AI models trained predominantly on data from high-income countries may perform poorly in the settings that matter most – missing signals that don't look like anything the system was trained on.
The equity gap runs through every layer:
Data coverage is uneven; many LMICs lack digitised health records
Connectivity is unreliable in the regions generating the earliest outbreak signals
Local health workers generating frontline observations often lack basic diagnostic infrastructure
An algorithm in Toronto or Sydney can only act on signals that reach it.
Progress is happening. EPIWATCH open-sourced its platform for use in LMICs, supported by a philanthropic grant from Ethereum co-founder Vitalik Buterin. A 2025 World Economic Forum initiative reported early pilot deployments of AI-assisted Mpox skin-image detection tools in African settings, enabling frontline workers to flag suspected cases from a mobile device.
These are meaningful steps. But the gap between what surveillance technology can do in London and what it can do in Kinshasa remains wide. Closing it requires investment in local laboratory capacity and digital health infrastructure – not just better algorithms. Without ground-level data, the most sophisticated AI platform generates silence.
What comes next
The current generation of AI surveillance tools is better at detecting known pathogens in well-connected settings than at catching genuinely novel threats in fragmented ones.
Hantavirus, Ebola, Mpox – these are known threats. The systems being built now need to catch the unknown ones too.
The RANGER (RApid Next Generation Sequencing for Effective Medical Response) project, a collaboration between Ginkgo Bioworks and the European Health and Digital Executive Agency funded through the EU4Health programme, is developing point-of-care metagenomic sequencing tools targeting rapid turnaround times on the order of hours. Unlike conventional diagnostics, metagenomic sequencing doesn't require knowing what you're looking for – it sequences everything and flags anything unusual, including pathogens that don't yet have a name.
A 2025 RAND Corporation analysis suggests that this kind of environmental sampling detected simulated outbreaks faster than wearable sensors or symptom-based reporting, strengthening the case for permanent multi-modal biosurveillance at airports, in wastewater networks, and in clinical settings.
The tools are arriving. AI platforms that scan 40 languages in real time, sequencing systems that don't need to know what they're looking for, open-source intelligence reaching into settings that traditional surveillance never could.
For the first time, the early warning system is starting to match the speed of the threat – and that's a very good thing for everyone.
Innovation highlights
🦠 Bacteria's secret codes, cracked. Antibiotic resistance genes are evolving faster than the databases designed to track them. ResLens is a new genomic AI model published in npj Antimicrobials and Resistance that detects resistance genes conventional tools miss – using transfer learning rather than building from scratch. It outperformed other deep learning tools on long-read DNA data and identified gene families that standard database-matching methods failed to catch entirely. Promising for screening, though researchers flag it still needs careful validation before clinical use.
🤰 A patch that watches over. A soft, wearable ultrasound patch developed at UC San Diego can continuously monitor a fetus and umbilical cord for hours – no sonographer required. Autonomous tracking algorithms follow the baby's movements in real time, catching fluctuations that standard snap-shot ultrasounds would miss. Tested across 62 pregnancies at UC San Diego and Oxford's John Radcliffe Hospital, it once detected an abnormality that led to an early delivery at 29 weeks.
🐾 Good boys, great medicine. Facility dogs – full-time, specially trained canine staff – are quietly becoming standard at children's hospitals across the US. Unlike volunteer therapy dogs, they work daily alongside clinical teams, supporting kids through painful procedures, motivating movement, and making wards feel less frightening. Research backs it up: even brief interactions lower cortisol, reduce pain scores, and bring down blood pressure
Company to watch
Perimeter launched in April 2026 as what it describes as the world's first integrated biosecurity infrastructure platform. The platform combines airport-based genomic surveillance, wastewater monitoring, and AI-driven threat characterization into a single continuously updated risk picture – currently deployed across eight US international airports and international nodes in Africa, the Middle East, and Ukraine.
Two recent milestones make it worth watching: in August 2025, the platform identified a novel H3N2 influenza subclade seven days before it appeared in any public repository – the strain went on to dominate globally. When Mpox Clade Ib rendered standard diagnostics ineffective in 2024, Perimeter designed and validated a replacement wastewater PCR assay within approximately three weeks. A company that has already demonstrated it can move faster than the threat.
Image source: Perimeter
Weird and wonderful
🧬 Helical slides are the new stairs. Most school science labs still look like they did decades ago: fixed benches, fume hoods, and a tired periodic table poster. At Institut Le Rosey, one of Switzerland’s most exclusive boarding schools, the newly completed Philo Science and Innovation Center takes a very different view.
Architect Bernard Tschumi has delivered a five-storey, ring-shaped building in Rolle, near Geneva, wrapped around a grand central atrium, with helical slides linking the floors. Inside, the program is built around student innovation: a Fabrication Lab, a Start-up Incubator Space, and a flexible Pitch Room that can shift between presentations and performances. Classrooms and laboratories fill the remaining levels, all oriented to the central void. It may be the only science centre where getting to class is the best part.
Image source: Yanko Design
Thank you for reading the Healthy Innovations newsletter!
Keep an eye out for next week’s issue, where I will highlight the healthcare innovations you need to know about.
Have a great week!
Alison ✨
P.S. If you enjoyed reading the Healthy Innovations newsletter, please subscribe so I know the content is valuable to you!
