NHMRC funding for innovative AI tech that will transform hospital data coding
Dr Ting Xia
We are excited to announce that a team of researchers led by Monash Addiction Research Centre (MARC) member Dr Ting Xia has been awarded a National Health and Medical Research Council (NHMRC) Ideas Grant!
Congratulations to the whole team, which includes MARC member Dr Tina Lam as a chief investigator with Dr Joanna Dipnall and Dr Jane Hayman.
The project aims to improve hospital emergency department (ED) data coding about the presence of opioids using an innovative form of AI language processing, which will enable automated and reliable surveillance of opioid-related harms.
“By developing the new technology to capture good quality, structured ED data from existing unstructured medical records, we will be able to monitor rates of opioid harm more closely and understand whether current policy interventions lead to their intended outcomes,” Dr Xia explained.
“It also has the potential to provide timely identification of unintended consequences or negative outcomes from policy changes, enabling more responsive and effective interventions,” she added.
“With recent changes in opioid policies and the drug market driving shifts in patterns of harm – such as the transition to illicit opioids or other drugs – this technology is well-positioned to capture these evolving trends.”
Opioid-related harms being missed in surveillance
Opioid-related mortality in Australia doubled in the past decade, leading to a range of policy changes to urgently address the growing harm from opioid use.
“Since many opioid-related harms, such as overdoses, are handled in EDs without patients being admitted to hospital, EDs are uniquely positioned to monitor drug harm and injuries,” Dr Xia said.
Currently, however, ED data remains underutilised to help researchers determine positive or negative outcomes from policy interventions because existing coding approaches lack the detail needed.
“This is partly due to the quality of the collected data,” Dr Xia explained. “For example, some ‘condition’ codes are often filled out incorrectly or not at all. About half of all ED opioid poisoning cases are not correctly documented.”
The project that has just received funding will seek to catch these undocumented cases by developing new technology to transform ED data collection processes.
“We know that a range of important details are recorded in the free-text sections of Electronic Health Records,” Dr Xia said.
For example, in the case of opioid poisoning, clinicians often write down specifics about the substances involved, such as “took Panadeine Forte,” “took methadone tablets,” or “intentional OD”.
“Natural Language Processing (NLP) is a cost-effective way to tap into this rich resource of unstructured text without adding extra work for clinicians or changing current practices,” Dr Xia said.
AI language processing to improve data coding
NLP is a form of AI that enables computer algorithms to understand and interpret everyday human language.
In the cases of opioid involvement in ED presentations, NLP algorithms can identify any missed examples of medical conditions using the free text areas of the Electronic Health Records.
The algorithms then accurately code the details in free text, making hospital data relating to opioids more accurate and detailed.
Once the new technology has been developed, more accurate, reliable and timely reporting on policy outcomes will be possible.
“This innovative approach to coding ED data opens the door to entirely new questions about policy performance, providing valuable insights to guide future policy decisions and prevention strategies,” Dr Xia said.
Congratulations to the whole team.
Find out more about MARC research about opioid prescribing in Australia and the policy impact: Enabling evidence-informed policy to address Australia's opioid crisis