Ventilator-Associated Pneumonia Prediction Models Based on AI: Scoping Review

Area of Science:

• Medical Informatics
• Clinical Prediction Models
• Artificial Intelligence in Healthcare

Background:

• Ventilator-associated pneumonia (VAP) is a significant complication of mechanical ventilation, impacting patient outcomes.
• Artificial intelligence (AI) offers advanced data mining capabilities for predicting VAP.
• Early identification of high-risk patients is crucial for effective VAP management.

Purpose of the Study:

• To review existing AI-based models for VAP prediction.
• To provide a reference for future clinical applications in identifying high-risk VAP groups.
• To synthesize current research on AI for VAP prediction.

Main Methods:

• A scoping review was conducted following PRISMA-ScR guidelines.
• Searches were performed across multiple databases including PubMed, Embase, and Web of Science.
• Data extraction and synthesis were conducted by two independent reviewers.

Main Results:

• Eleven studies utilizing AI for VAP prediction were included.
• All reviewed studies focused on VAP occurrence prediction, excluding prognosis.
• Machine learning, particularly random forest, was the predominant algorithm; text data was used, but not imaging data. Public databases were common data sources, with most studies having sample sizes under 1000. Deep learning and large language models were not employed. All studies reported only internal validation, lacking real-world implementation strategies.

Conclusions:

• AI models demonstrate superior predictive performance for VAP compared to traditional methods.
• AI is poised to become an essential tool for VAP risk prediction.
• Further research is needed to address the clinical implementation and practical guidance for AI-driven VAP prediction models.