Association for Diagnostics & Laboratory Medicine

Dr. Stephen Master, vice chair for innovation in the department of pathology and laboratory medicine at Children’s Hospital of Philadelphia, discussed examples of variables that can influence the accuracy of artificial intelligence (AI) healthcare tools in laboratory medicine.

One variable is the instrument or manufacturer used to generate test results. Many labs still produce different results depending on the instrument and manufacturer. This can influence how AI healthcare tools report test results.

Another variable is the demographic data that an AI algorithm was trained with. If it was trained with data from adult patients, can it adequately address the test results of pediatric patients?

Given these and many more variables, laboratories need transparency from developers on the data used to train AI healthcare tools.