Machine learning improves prediction of pressure ulcers

What is it about?

We show that machine learning methods can classify at risk patients for pressure ulcers in the intensive care unit (ICU) by using electronic health information (EHI) captured in chart records. We improve on previous methods such as the Braden and Waterlow scoring systems by proposing a model which could automate patient risk classification by using their chart data.

Featured Image

Why is it important?

Pressure ulcers are a significant health issue that are particularly difficult in ICU settings. They can lead to open wounds which present an additional opportunity for infection and sepsis among already ill patients. This is an important step forward because the nature of pressure ulcers means prevention and care is a major hospital (ICU) complication and concern everywhere. If we can predict patients at risk of developing a pressure ulcer, then we can take preemptive preventative measures to improve patient outcomes and save scarce hospital resources.