> The surveillance system doesn't automatically flag health concerns – often scientists dive into the data once their suspicion is raised, as it was in this case. The outbreak only became apparent in the data once researchers defined more closely what they were looking for and focused on one age group.
I'd be surprised if there wasn't some kind of outlier detection method that could automatically flag a sudden increase in unusual cases.
The web site has a lot of papers talking about that. They do detect spikes in defined syndromes.
But for stuff like this that isn’t a defined syndrome and isn’t really captured well at the hospital it’s not an outlier until scientists do a lot of analysis.
The ETL for 3200 emergency rooms worth of data must be a nightmare. It would have to be largely unstructured upon ingestion, and I can’t imagine normalising it enough for this sort of automated outlier detection would be possible.
I'd be surprised if there wasn't some kind of outlier detection method that could automatically flag a sudden increase in unusual cases.