New machine learning models developed by University of South Australia (UniSA) researchers could help clinicians identify when patients can successfully stop long-term antidepressant use.