Automated risk assessment tool for pregnancy care

Area of Science:

• Obstetrics and Gynecology
• Medical Informatics
• Machine Learning in Healthcare

Background:

• Clinical decision support systems enhance medical care quality by assisting in complex case evaluation.
• Accurate assessment of pregnancy criticality is vital, requiring interpretation of multiple maternal and fetal parameters.
• Existing methods may benefit from advanced analytical tools for timely risk stratification.

Purpose of the Study:

• To propose a machine learning approach for the early determination of pregnancy risk categories.
• To leverage classification and regression trees for analyzing multivariate obstetric data.
• To illustrate the utility of this approach through a proof-of-concept application.

Main Methods:

• Utilized a machine learning approach to analyze patterns in clinical parameters.
• Employed classification and regression trees (CART) for multivariate problem-solving in obstetric care.
• Developed and presented an application use case for early risk determination.

Main Results:

• Demonstrated the effectiveness of machine learning, specifically CART, in identifying pregnancy risk categories.
• Highlighted the interpretability of the decision-making process and parameter importance through tree visualization.
• Successfully applied the model in a proof-of-concept scenario.

Conclusions:

• Machine learning offers a valuable tool for early and accurate pregnancy risk assessment.
• Classification and regression trees provide transparent and interpretable insights into obstetric decision-making.
• This approach can significantly aid healthcare providers in managing high-risk pregnancies.