Antibody Identification and Software Processes

Processes and Software Building—Part 18

The blood bank is one of the last bastions of manual testing, especially for complex procedures.  Basic antigen typing, direct antiglobulin tests, and antibody screening adapt well for routine automation, but the actual antibody identification requires multiple methods and requires considerable experience to complete. At least today, a machine cannot replace the blood banker!!

One could take the approach to perform antibody panels in an automated system and record the results.  However, these are worthless unless you have the actual panel in hand.  Artificial intelligence to analyze the panel results is limited.  You still have to manually review everything.  In Medinfo, we could also scan the manually performed panels and add them to the record.

While I was at HMC, some of the hospital transfusion services elected to run panels automatically and then review the results manually.  We prepared a contingency to record the panel results but these by themselves could only be used with the actual red cell panels in use.

Examples of 11-R and 15-cell panel result importation follow:

What we absolutely had to do was enter the antibody specificity in a way that it could trigger our algorithms, i.e. our rules-based system of RBC allocation according the permissible blood types.  Each antibody specificity and each antigen typing had its own unique codeIt could not be entered free text!  Workflows follow:

Next, the antibody specificity test would trigger antigen typings for the correspond antigen and/or closely related antigens:

Based on the antibody specificity and antigen typings, a whole series of rules for routine and emergency-mode release were triggered.  This will be discussed in a subsequent post.

To Be Continued: