588 followers
@iddux [caveat: I'm not a doctor, and certainly not a radiologist] This strikes me as a reasonable approach! Would you use billing data as a starting point, or the CXR images themselves? If the latter, a computer could maybe ID specifically the viral bila