New research from the University of South Australia (UniSA) suggests artificial intelligence (AI) could soon give general practitioners the confidence to know when patients can safely stop long-term antidepressant use.
The study, presented this week at MedInfo 2025, an international conference on digital health and informatics, found that machine learning models can accurately predict which patients are most likely to withdraw successfully from antidepressant medication.
Researchers analysed dispensing data from the Pharmaceutical Benefits Scheme (PBS), tracking 100,000 patients who had been prescribed antidepressants over a 10-year period. By applying AI to this large dataset, they identified the most successful cases of deprescription, defined as stopping antidepressant use for at least one year following more than 12 months of treatment.
The results are promising. One AI model that assessed final prescription records achieved an accuracy rate of 81 per cent, while another that tracked patients from their first prescription, including dose reductions and outcomes, achieved 90 per cent accuracy.
Dr Lasantha Ranwala, a medical practitioner, AI researcher and UniSA PhD candidate, said the tool could help clinicians manage the delicate balance between therapeutic benefits and risks.
“Healthcare providers are often reluctant to cease antidepressant prescriptions due to concerns about withdrawal effects, making it difficult for doctors to know who can safely discontinue treatment,” Dr Ranwala said.
“By applying AI to the PBS database, we have identified patterns linked to successful withdrawal, forecasting which patients are most likely to succeed when taking them off antidepressants.”
While antidepressants can be life changing, prolonged use is not without consequences. Side effects may include weight gain, sexual dysfunction and cardiac issues. At the same time, half of all patients experience withdrawal effects when they stop taking the medication, ranging from dizziness and insomnia to flu-like symptoms.
With antidepressant use soaring globally — Australia, Iceland, Portugal, Canada and the UK record the highest consumption rates — UniSA researchers say the findings could provide doctors with a valuable decision-support tool.
Associate Professor Andre Andrade, a co-author of the study, said the ability to predict successful withdrawal could give doctors greater confidence to act.
“The most accurate model was the one that offered a more nuanced picture of deprescription attempts, better reflecting patient experiences,” Assoc Prof Andrade said.
“This data is passively collected, underused by medical professionals and a good candidate for AI use.”
The team will now focus on improving the technology’s accuracy and making it more user-friendly in clinical practice. They also hope to explore similar applications of AI to optimise medicine use more broadly.
In Australia, antidepressants were dispensed to 14 per cent of the population in 2023–24, according to OECD data. The highest relative increase in long-term use was seen among young people aged 10–24, where use more than doubled in a decade. Across all age groups, the average treatment episode also grew longer, with young people showing the most significant rise.
Researchers believe AI could play a critical role in helping health systems adapt to these trends by supporting safer prescribing and deprescribing decisions.
The study, ‘Predicting Antidepressant Deprescription with Machine Learning Using Administrative Data,’ highlights the potential for digital health tools to improve outcomes for patients and ease pressure on GPs managing long-term antidepressant care.
