Muscat – A research team in Omani developed a health system powered by artificial intelligence that can predict childhood obesity with 97% accuracy, and marked the region first in the use of Health Internet Things (IOHT) and machine learning to combat the growing threat of obesity.
The project is led by Navira Al Rushdi, a senior lecturer in the Faculty of Engineering at the National University of Science and Technology and funded by the Ministry of Higher Education, Research and Innovation – combining IOHT devices, health sensors and electronic medical records to support early prediction, prevention and management of children.
Nabila said the system analyzes biometric indicators and electronic health records, particularly those collected from birth to age 2, to predict the risk of childhood obesity. “By analyzing biometric indicators and electronic health records, especially those collected from birth to age 2, we have been able to predict the risk of childhood obesity with incredible accuracy,” she said.
Of several machine learning models, including Random Forest, Support Vector Machines and XGBoost, logistic regression was the most accurate, predicting the risk of childhood obesity by age 10 with an accuracy of 97.09%. The team also develops mobile and web applications, allowing users to monitor their calorie intake, weight, sleep and physical activity, and receive customized weight loss advice and health tips.
The platform's system evaluates factors such as age, gender, BMI, nutritional habits, and physical activity to provide real-time, evidence-based recommendations for both patients and healthcare professionals.
Based on the findings, the research team recommends integrating smart healthcare platforms into national health programs to enhance early detection and interventions, particularly for children. It also encourages greater investment in AI-powered healthcare, promoting forecasting and personalized medicine across Oman and beyond.
The team, overseen by Dr. Varghese MJ, has submitted a scientific manuscript detailing the findings in a peer-reviewed journal, awaiting publication.
