Finished! Looks like this project is out of data at the moment!
This project is now complete! Many thanks to everyone who participated. See the results
FAQ
Where has the data come from?
The Infections in Oxfordshire Research Database (IORD) is a continually-updating data store containing anonymised Electronic Health Record data from all patient admissions to the Oxford University Hospitals since 1998, linked with microbiology and biochemistry/haematology laboratory results. It currently comprises over a billion records across ~700 data fields, and is known to cover multiple periods of change in the hospital computer and laboratory systems. IORD has Research Ethics Committee, and Health Research Authority and Confidentiality Advisory Group approval as a generic electronic research database without individual patient consent (14/SC/1069, ECC5-017(A)/2009).
How many sudden changes should I expect to see in a single image?
It really varies from one image to another. Some will have no sudden changes, others may have lots, though we would typically expect to see no more than ten in a single image.
How big does a change need to be in order to qualify?
Bigger changes are easier to spot, but we are interested in small changes too, and there isn't a specific minimum size below which changes should be ignored. It is more about identifying points that look suspicious, and this will depend on what the rest of the data in the graph looks like. You can draw a yellow line if you are not sure.
What do I do with outliers?
Please mark any outliers as well, by putting a line through them. While outliers may seem to be different to the other types of "sustained" changes, you can think of them as two changes in quick succession, and we are interested in all cases of "suspicious" behaviour, including short-lived ones.