ORIENTATE: automated machine learning classifiers for oral health prediction and research

Area of Science:

• Healthcare Informatics
• Clinical Decision Support Systems
• Machine Learning in Medicine

Background:

• Data-driven methods are crucial in healthcare but adoption is limited by a skills gap.
• Clinical practitioners often lack the expertise to develop and interpret machine learning models.
• ORIENTATE software addresses this gap by automating machine learning classification.

Purpose of the Study:

• To introduce ORIENTATE, a user-friendly software for automated machine learning classification in clinical practice.
• To enable clinical practitioners without specific technical skills to build and interpret predictive models.
• To facilitate statistical inference and complement classical studies using machine learning outputs.

Main Methods:

• ORIENTATE automates feature selection and target variable identification.
• The software generates, cross-validates, and evaluates multiple classification models.
• It includes a custom feature selection algorithm for systematic predictor searches.
• Comprehensive reports with global interpretation methods and prediction interfaces are generated.

Main Results:

• ORIENTATE was applied to a dataset of children undergoing deep sedation.
• The software achieved a high prediction accuracy for the need for a second sedation (F1 score 0.83, ROC AUC 0.92).
• Eight predictive factors were identified, with the number of teeth requiring pulpal treatment at the first sedation being a key predictor.

Conclusions:

• ORIENTATE empowers researchers and clinicians to leverage machine learning for predictive and inferential tasks.
• The software facilitates the application of machine learning classification for preventive healthcare.
• It serves as a valuable complement to traditional statistical studies, enhancing data interpretation.