Since we are making predictions at the end of each admission, it would be useful to have an m-by-n matrix of m admissions by n features, where each feature represents the last value before discharge. It may also be useful to add a time constraint, e.g. the value must occur within 72 hrs of discharge, etc.
Since we are making predictions at the end of each admission, it would be useful to have an m-by-n matrix of m admissions by n features, where each feature represents the last value before discharge. It may also be useful to add a time constraint, e.g. the value must occur within 72 hrs of discharge, etc.