Diagnosis can be defined as the identification of a disease in an individual. In contrast to screening, diagnosis may happen at any time during the course of the disease. Diagnosis is important for almost every disease because it dictates how the signs or symptoms (and the underlying disease) should be treated. The exception occurs when diseases have no effective treatment, or when differentiating between diseases does not change the treatment.

A common use of machine learning in diagnosis problems is to identify potential causes of underlying disease in the face of a mysterious symptom, for example, abdominal pain. In contrast, building a machine learning model to differentiate between different types of psychiatric personality disorders may be of limited efficacy, since personality disorders are difficult to treat effectively.