In a recent study, a new algorithm achieved good performance for predicting which hospitalized patients will develop acute kidney injury requiring dialysis.
"A machine learning model using admission features had a good performance for prediction of dialysis need," said study co-author Lili Chan from Mount Sinai Health System in the US.
Models like this are potentially useful for resource allocation and planning during future Covid-19 surges. We are in the process of deploying this model into our healthcare systems to help clinicians better care for their patients," Chan added.
According to the researchers, preliminary reports indicate that acute kidney injury is common in patients with Covid-19.
Using data from more than 3,000 hospitalised patients with Covid-19, investigators trained a model based on machine learning, a type of artificial intelligence, to predict AKI that requires dialysis.
Only information gathered within the first 48 hours of admission was included, so predictions could be made when patients were admitted.
The model demonstrated high accuracy (AUC of 0.79), and features that were important for prediction included blood levels of creatinine and potassium, age, and vital signs of heart rate and oxygen saturation.
The research is scheduled to be presented online during ASN Kidney Week 2020 Reimagined October 19-25.