Artificial intelligence-enabled obesity prediction: A systematic review of cohort data analysis

Area of Science:

• Health Informatics
• Artificial Intelligence in Healthcare
• Predictive Analytics

Background:

• Obesity is a leading global cause of death, increasing risks for type 2 diabetes and cardiovascular disease.
• Early obesity risk identification is crucial for preventative interventions.
• Developing comprehensive AI-based obesity screening tools requires extensive cohort data.

Purpose of the Study:

• To review and critically appraise existing studies on AI predictions for obesity in cohort studies.
• To identify common AI algorithms, data types, and performance metrics used in obesity prediction.
• To inform dietitians and researchers on AI applications for obesity-related predictive modeling.

Main Methods:

• A systematic literature review of 6,351 articles up to March 2024.
• Focus on AI predictions for obesity within cohort studies.
• Critical appraisal of 10 selected studies using the JBI checklist.

Main Results:

• Ten studies involving 411,580 participants were analyzed, with varying cohort lengths and sizes.
• Demographic and biomarker data were most frequently utilized.
• Machine learning, particularly supervised learning techniques like random forest, linear regression, and gradient boosting, dominated the studies, with high performance reported for models like k-means and artificial neural networks.

Conclusions:

• AI algorithms demonstrate potential for obesity prediction.
• Further research is necessary to evaluate AI's effectiveness in analyzing obesity data and explore advanced AI methods.
• This review serves as a resource for developing AI-driven predictive models and clinical decision support systems.