AI to help hospital discharge

A major new $2.9 million digital health project led by Flinders University will target the dosage and failures of prescription drugs used by patients at home after they leave hospital.

The ‘AutoMedic’ project aims to create a more streamlined and accurate electronic record of prescription medicine used by patients discharged from six South Australian public hospitals to better manage their recovery and longer-term health outcomes.

The trial will provide a useful model for a “smart, scalable solution to detect and resolve medicine harm,” says the Chief Investigator of the $2.9 million Australian Government Medical Research Future Fund National Critical Research Infrastructure Grant project, Associate Professor of Digital Health Niranjan Bidargaddi, from the College of Medicine and Public Health.

SA Government support includes key partners in the project – SA Pharmacy and Digital Health SA – with other academic, industry and government partners.

In image of Associate Professor Niranjan Bidargaddi
Associate Professor Niranjan Bidargaddi is a computer science engineer and e-health expert.

Nearly 17 million Australians are on at least one prescription medicine a year, with some high-needs patients managing multiple drugs for various co-occurring conditions. As many as 90% of hospitalised patients could have medicine-related problems after leaving hospital, including use of inappropriate drugs, dosage or duration of treatment – particularly when using high-risk drugs for co-morbidities.

With about 300,000 public hospital presentations every year, more accurate e-health solutions will create an AI-enhanced medicine review for individual patients, supporting  hospital clinical pharmacists to give advice or alternative prescriptions in a more timely manner before discharge.

“This new e-health system will help identify patients most at risk of adverse events due to medicine errors when in hospital and put GPs and pharmacies in a position to intervene and assist and even prevent such problems, which can lead to re-hospital readmissions,” says health informatics expert Associate Professor Bidargaddi, who leads a Digital Health Lab Flinders University.

“As well as improving medicine safety and precision, AutoMedic will increase efficiencies and cost savings in the State hospital agencies, notably SA Pharmacy, while creating a model to roll out nationally.

“Our research team, comprising national leaders in clinical care, algorithm development as well as government, academic and industry partnerships, has proven experience in delivering large-scale research projects in healthcare systems – including applying AI to My Health Record data.”

AutoMedic will be designed to intelligently become smarter over time, ranking patients based on the importance and urgency of the review required, with potential harm caused by a medicine regimen will be ranked higher than low-impact medicine regimens, he says.

SA Pharmacy Executive Director Richard Marotti says each day hospital pharmacists spend a significant amount of time prioritising patients for review and collecting information to reconcile a patient’s medication list.

“A technology enabled tool to assist in the reconciliation and identification of medicines risk, that is inclusive of a patient’s pre-hospital medicines, will greatly improve the timely delivery of clinical pharmacy services, prevent medication errors and support safer use of medicines overall,” Mr Marotti says.

The new system will work in tandem with the federal My Health Record platform, which collates prescription, imaging and pathology data from Medicare-funded programs, a new single electronic medical record (‘Sunrise AllScripts’) operating at public hospitals in SA.

The new project also involves other researchers from Flinders University, the Australian Institute for Machine Learning (AIML) and the University of Adelaide, UniSA and the SA Health and Medical Research Institute (SAHMRI).

Posted in
College of Medicine and Public Health Flinders Digital Health Research Centre Research