Reviewed by Lexie CornerApr 10 2024
According to the US Agency for Healthcare Research in Quality, bedsores, often referred to as pressure injuries, are the fastest-growing hospital-acquired condition. As a result, they are now the second most prevalent cause of medical negligence lawsuits in the country.
Although most hospital-acquired pressure injuries are generally avoidable, roughly 2.5 million people in the United States suffer a pressure injury in acute care facilities each year, with 60,000 dying. The yearly cost for US health systems to handle the acute requirements of patients with pressure injuries during hospitalization exceeds $26 billion. Still, pressure injuries have received little attention as a public health concern.
Researchers from USC, Johns Hopkins University, and University Hospitals Cleveland Medical Center collaborated to create a novel model that uses machine learning to anticipate potential pressure injuries and better coordinate labor-intensive patient care.
The new risk-assessment approach, published in BMJ Open, improves prediction accuracy to more than 74 %, representing a more than 20 % increase over existing techniques.
Common techniques and recommendations for preventing pressure injuries are time-consuming and demanding on nurses at the bedside. The paper-based Braden Scale, the industry standard for forecasting the risk of pressure injury, has remained unchanged since its inception in the 1980s and has a 54 % accuracy rate, according to the study.
Saving Time, Costs and Lives
According to the researchers, predictive analytics has the potential to reduce the strain on nurses and frontline healthcare personnel by automating some aspects of the risk assessment process.
Currently, acute care providers are required to do a skin check and risk assessment for pressure injury at admission and every 12 to 24 hours thereafter, using a standardized instrument such as the Braden Scale, which examines mobility, cognition, nutrition, and incontinence management.
Pressure injury prevention is a costly protocol to implement on a daily basis, and the existing tool for predicting pressure injuries is barely better than a coin flip. We thought, there’s got to be a better way of doing this. The question became, could a computer do these risk assessments better than the nurses themselves at the bedside?
William Padula, Department of Pharmaceutical and Health Economics, USC Alfred E. Mann School of Pharmacy and Pharmaceutical Sciences
The team’s predictive algorithm improves economic efficiency and generates significant savings. A risk assessment can take five to 15 minutes for each patient, adding up to 250 labor hours per day in a single 500-bed facility, or 30,000 to 90,000 labor hours per year.
This data could help hospitals conserve resources within a critical period of patient vulnerability of hospital-acquired pressure injury that is not reimbursed by US Medicare.
Peter Pronovost, Chief Quality and Transformation Officer, University Hospitals Cleveland Medical Center
The research also promotes advances in health equity. Existing tools do not account for race, skin color, or age.
Padula added, “If you can’t see bruising on a patient’s skin because they are Black, Hispanic, or Asian, then you’re not going to identify the greater risk factors they face as quickly. Machine-learning methods are not biased by what we see in the sun’s light. This allows us to improve equity of the delivery of healthcare when it comes to the prevention of these conditions that disparately affect underrepresented minorities.”
Artificial Intelligence Methods
Using machine-learning techniques, researchers examined the electronic health records of over 35,000 hospitalizations at two academic hospitals over five years to examine variations in pressure injury risk. They examined variables such as admission diagnostic codes, prescription drugs, lab orders, and other parameters closely related to pressure injury risk factors.
AI-based early detection significantly outperforms standard of care. Hospitals can use this to initiate a quality-improvement program for pressure-injury prevention that improves outcomes and significantly lowers the burden on nursing from current monitoring approaches. Further, they can customize the algorithm to patient-specific variation by facility.
Suchi Saria, John C. Malone Endowed Chair and AI Professor, Johns Hopkins University
The research used machine-learning approaches such as random forests and neural networks to refine the exact weights of those factors on the changes and risk of a pressure injury instance, resulting in the final model.
They also found several prescription drugs—beta blockers, electrolytes, phosphate replacement, zinc replacement, erythropoietin stimulating drugs, thiazide/diuretics, and vasopressors—that alter patients’ risk of pressure injury.
David Armstrong, professor of surgery at Keck School of Medicine of USC, added, “This is, to our knowledge, the most advanced methodological study to our knowledge to use artificial intelligence to help better detect pressure injuries.”
This research was funded by a National Institutes of Health grant (KL2 TR001854, PI: Padula).
Journal Reference:
Padula, W. V., et. al. (2024) Predicting pressure injury risk in hospitalized patients using machine learning with electronic health records: A US multilevel cohort study. BMJ Open. doi:10.1136/bmjopen-2023-082540