Machine Learning for Estimating Cardiorespiratory Fitness in Patients With Obesity: Protocol for a Retrospective and...

Area of Science:

• Cardiology
• Exercise Physiology
• Medical Informatics

Background:

• Cardiorespiratory fitness (CRF) is crucial for predicting health outcomes in individuals with obesity.
• Maximal oxygen uptake (VO2max) testing is the gold standard for CRF assessment but is impractical for widespread use.
• Existing predictive models for CRF often perform poorly in obese populations due to data limitations.

Purpose of the Study:

• To develop and validate a clinically relevant machine learning (ML) model for estimating CRF in adults with severe obesity.
• To compare the ML model's performance against direct VO2max measurements.
• To create an accessible web application for the ML model.

Main Methods:

• Utilizing a retrospective dataset of VO2max tests and clinical parameters from adult patients with severe obesity (BMI ≥40.0 kg/m2 or 35.0-39.9 kg/m2 with comorbidities).
• Developing a machine learning model using routinely collected clinical measures.
• Prospective validation comparing ML estimations against direct VO2max measurements.

Main Results:

• Over 2623 VO2max tests and clinical data from 1279 adults with severe obesity were collected between 2013-2025.
• The ML model aims to provide a practical and accurate estimation of CRF.
• First scientific publication of the ML model is anticipated in 2026.

Conclusions:

• The developed ML model offers a cost-effective tool for CRF estimation in individuals with obesity.
• This initiative represents the first ML-based CRF estimation using a clinical database in Norway for this population.
• The project has significant societal value and national/international relevance for healthcare and patient outcomes.