Precision in Prevention: Tailoring Single-Use Negative Pressure Wound Therapy Utilization Through Artificial...

Area of Science:

• Medical Informatics
• Surgical Outcomes Research
• Health Economics

Background:

• Surgical site complications (SSCs) are common, preventable hospital-acquired conditions.
• Single-use negative pressure wound therapy (sNPWT) effectively reduces SSC rates.
• Optimizing sNPWT allocation is crucial for value-based care, balancing clinical and financial outcomes.

Purpose of the Study:

• To develop and validate machine learning models for predicting pre-operative SSC risk.
• To identify key predictors of SSCs across various surgical specialties.
• To simulate the cost-effectiveness of sNPWT based on individualized risk assessments.

Main Methods:

• Retrospective analysis of the Premier Healthcare Database (2017-2021) for 10 open procedures.
• Development of machine learning models using training and validation data sets.
• Feature importance evaluation to identify SSC predictors and cost-effectiveness simulations.

Main Results:

• Machine learning models achieved good predictive performance (average AUC of 76%).
• Key predictors included patient age, obesity, and procedure urgency.
• Simulations revealed significant variability in sNPWT cost-effectiveness across procedure types.

Conclusions:

• Machine learning accurately predicts patient SSC risk, enabling targeted sNPWT utilization.
• A data-driven, personalized approach optimizes both clinical efficacy and financial efficiency.
• Strategic sNPWT allocation based on risk assessment enhances value-based care.