Even if we had well defined schemas/data models, and profiles thereof that guide the way data is captured,
it is almost never the case that the (patient’s) instance data is perfect.
In fact, imperfection can arise from multiple sources: missing data, uncertainty, confidence/error ranges, vagueness, etc.
(see e.g. https://www.researchgate.net/figure/different-kinds-of-information-imperfection-18_fig2_257211032)
Neither data elements nor the models processing them should ignore this imperfection, otherwise
leading to a scenario that is commonly referred to as ‘garbage in-garbage out’
Options:
(1) Do not execute .unless all data is available and valid(ated)
(2) Look for another that can run on the available data.alternatives (including humans)
(3) Have an operational solution with order sets guaranteeing dataEnsure that the data is present implementing processes upstream.
(4) Order Acquire the missing data .
(5) Run a range of values for missing data and see if outcome changes.