You’re in hospital: should you stay? Should you leave? What’s your risk of dying?
By mining electronic health records, researchers at Macquarie University believe they can help improve decision making by health professionals.
Dr Blanca Gallego Luxan is investigating using hospital information and state health and death registries to fill gaps in patient care – whether due to discontinuity of care, lack of information on a condition, or simply the limits of what humans can predict.
She’s developing models to predict the likelihood of a patient staying in hospital, being discharged, readmitted, or dying within a certain time-frame. The patient’s ‘forecast’ is continually updated with the results of each of their medical tests.
Blanca has shown that the models work – particularly for predicting death in the short-term (within one week). Now the focus is on understanding how clinicians will use the models, their safety in clinics, and gaining ethics approval.
If a patient is likely to be readmitted or become sicker, it may prompt interventions or discussions on end of life choices, Blanca says.
“Humans are limited in their knowledge – there will be some cases where doctors just can’t predict what will happen, and they can’t be present with a patient 24/7.
“These models will help give staff the most comprehensive picture.”
And this project is just the start – they’re also working with Stanford University on decision-support tools for health professionals who are uncertain on the best treatment to prescribe for patients with complications.
This story was originally published Stories of Australian Science and is republished here with permission.