@OWMorgan

Director Pandemic and Epidemic Intelligence Systems @WHO #WHOPandmicHub #globalhealth #surveillance #riskassessment #healthemergencies #outbreaks

 Berlin, Germany
Joined January 2018
  • Tweets1.8K
  • Following214
  • Followers3.6K
203Media
If an AI study landed on your ethics committee tomorrow, could it review the algorithm? In Tanzania, researchers asked 10 committees. The governance was solid. The AI-specific capacity was not: judging algorithms, vetting big datasets, monitoring after approval.
1
36
"Just pilot it, train staff, embed it, move on." That is how we roll out most health innovations. A new analysis argues AI breaks that model. It can be opaque, and it drifts as data and vendors change. Adoption is not the finish line. It is the start of oversight.
89
In 2024, more than 60% of US local health departments reported no dedicated informatics staff. You cannot fix a data pipeline no one on staff can change. The workforce is a layer of AI readiness, not an afterthought. aipublichealth.substack.com/…
96
~250,000 adverse safety events hit US healthcare every year. More than 95% go undetected. A new tool in PLOS Digital Health goes after that gap, without handing the final call to a black box. 🧵
1
108
more replies
Tested on 300 paediatric cardiac arrest charts for adrenaline dosing errors: 97.9% accuracy on overdoses 91.6% on dangerous delays under 1 minute per chart It beat LLM-only baselines, and most misses came from data extraction, not the logic.
1
31
The catch: one drug, one setting, and a hand-built rule set for each new use case. But pairing model-based extraction with rule-based decisions is a workable answer to trust in clinical AI. Full digest: aipublichealth.substack.com/… #APHIDigest
23
1/ Most public health agencies are not ready for AI. Readiness isn't a model you buy. It's four layers of infrastructure, built in order. A short thread.
2
1
159
more replies
5/ Layer 4, partnerships. Where does capability sit when the contract ends? A vendor can leave, and the gap it filled comes straight back.
1
22
6/ The layers compound. Each dataset, standard, and skilled hire makes the next one worth more. Skip them, and every project starts from scratch. Full article: aipublichealth.substack.com/… #APHIArticle
21
This week's AI and public health journal digest is out. 6 papers on where AI is actually landing in health: a safety-event detector, an ambient scribe, and a hard look at whether AI is 'just another innovation'. Full digest: aipublichealth.substack.com/… #APHIDigest
93
A perfect sepsis alert does nothing in a clinic with no antibiotics, no fluids, and no power. A prediction with nowhere to act is worthless. AI readiness is about the whole system, not the model.
84
Buying a capable AI model does not make a health agency ready to use it. A model is only as good as what feeds it. Late data, no shared identifier, no one on staff to reshape the system: the tool inherits every gap beneath it.
1
103
More than half of sub-Saharan health facilities have no reliable electricity. About a third of the continent uses the internet, against three-quarters globally. These are not problems a better AI model can solve. aipublichealth.substack.com/…
1
86
1.7%. That is the share of 33,255 infectious disease modelling papers (44 years, mapped by an LLM pipeline) ever cited in a policy document. We are modelling more than ever and reaching decisions less. Output stays concentrated in high-income countries.
1
4
112
A health department can buy an excellent AI model and get nothing back. The data arrive late, in a format it cannot read, from a system no one can change. AI readiness isn't a model. It's four layers of infrastructure. aipublichealth.substack.com/… #APHIArticle
85
AI in the clinic is not adopted because the evidence demands it. It is adopted because of how it is sold. Same IVF embryo-selection tools: UK marketing calls them inevitable and slides "patient choice" into paid add-ons, while Dutch clinics hold the evidence line.
1
98
Load more