Data science and machine learning in anesthesiology.

Area of Science:

• Anesthesiology
• Artificial Intelligence
• Data Science

Background:

• Classical research methods in anesthesiology are inference-based.
• Machine learning (ML) offers a prediction-focused approach.
• ML applications are increasingly vital in medical research.

Purpose of the Study:

• To introduce basic machine learning concepts for anesthesiology researchers.
• To provide practical guidance on software and hardware for ML applications.
• To review successful ML applications in anesthesiology using electronic health record (EHR) data.

Main Methods:

• Focus on supervised ML applied to EHR data.
• Comparison of various ML models against logistic regression.
• Discussion of ML workflows for prediction tasks.

Main Results:

• ML excels at handling complex interactions and maximizing predictive performance.
• EHR data offers rich information for uncovering correlations at low cost.
• ML applications can identify previously unknown patterns in patient data.

Conclusions:

• ML is revolutionizing anesthesiology research through predictive modeling.
• While causality can be challenging, ML in EHRs holds significant potential.
• This article serves as a foundational guide for novice ML researchers in anesthesiology.