Standard psychological tests used to identify people at high risk of suicide missed the vast majority of people who would go on to kill themselves, the Royal Commission into Defence and Veteran Suicide has heard.
The executive director of Metro North Mental Health in Queensland, Dr Kathryn Turner, told the commission her realisation "we needed to do things differently" had prompted a major change in approach at the Gold Coast Mental Health Service.
Dr Turner, a psychiatrist, said she and her colleagues recognised the system was failing the large number of people who suicided because it was based on the assumption people with suicide were suffering a mental illness.
"Through my whole career, the focus had been on risk assessment ... and yet we knew we were seeing people who were dying by suicide that did not fit into that group," Dr Turner said.
"Our job was identifying people with mental illness ... but what we found among those people dying by suicide, only a very small number would be in that group."
Dr Turner said in 2016 the Gold Coast Mental Health Service launched its "Journey to Zero" program.
"Our clinicians were given permission to ... connect with people who don't necessarily have a mental illness but are in crisis and distress," Dr Turner said.
"Often we thought our jobs were to identify someone with mental illness ... Whereas the shift was that actually people come in with distress and suicide ideation without a diagnosis.
"They don't necessarily have a serious mental illness but are still in need of care and connection to ongoing support."
The commission heard the move away from the "tick a box" approach where patients were assessed as low, medium or high risk also involved reaching out to families who were dealing with the devastating impact of a suicide.
"Its (also) about healing ... we get much better learnings and improvement with this than we ever got with the old way we went about," she said.
She said a review of the new treatment approach had shown a 35 per cent reduction in suicide re-attempts.
While that was "encouraging", the hope was the approach could be further refined.
One potentially powerful tool to identify future suicide predictors, she said, may be the use of machine learning.
Several international studies were currently analysing patient health records to establish what red flags to look out for, dating back as far as 15 years.
Lifeline 13 11 14
Open Arms 1800 011 046
Australian Associated Press