Assume all images are COVID positive?

[ieee8023]

If the patient is not admitted yet (positive admission offset) should we assume they are COVID positive? Possibly just not testing positive at the moment?

[liob]

This is an interesting question. Our strategy is to represent the data as faithfully as possible. We thought about different representations:

• binary: COVID yes / no
• three states: COVID no / likely / yes

However, in the end, we decided against such a representation and opted for the admission offset. The idea was to let the data analyst decide which representation best fits for their specific research hypothesis. Unfortunately, this puts you in the position to define criteria.

If I had to decide, I would declare:

• negative: admission > +14
• positive: admission < +7
• unknown: admission <= +14 & admission >= +7

We are currently working on making the first positive test date available as an offset. Would this help you in this regard?