Use of AI to predict risk of HAIs

Editor's Note

This study led by researchers at the Mayo Clinic Arizona finds that a healthcare associated infection (HAI) risk prediction artificial intelligence (AI) model can estimate individualized risk of infection by taking into account a patient’s clinical features and features of similar patients.

The model was trained on 38,327 unique hospitalizations, and a distinct model for surgical site infection prediction was trained on 18,609 hospitalizations.

The model outperformed all single-modality  and length-of-stay models, and cost-effective analysis showed that the model was better than the standard length of stay model strategy on the basis of lower mean costs ($1,651 vs $1,915).

The researchers concluded that the model could allow prevention or earlier detection of HAIs, which could decrease hospital length of stay, associated mortality, and healthcare costs.