An artificial intelligence based algorithm for verifying clinical biochemistry tests

In medical laboratories, an automatic verification system refers to a process where computer-based algorithms automatically perform actions on laboratory test results based on predefined rules, without the need for manual intervention by laboratory personnel, or through more advanced algorithms for result evaluation. In this thesis, selected clinical biochemistry tests will be verified using both a rule-based automatic verification system and a machine learning-based automatic verification system. During this study, work was conducted on four different machine learning methods are examined, and the automatic verification rates for the selected tests are observed. Two separate datasets were chosen for this study, one used for training the machine learning system and the third executed in real-time with comparisons made against approvals by clinical experts. As a result, an exemplary structure for the automatic verification system of clinical biochemistry tests was created.