PP413 - ARTIFICIAL NEURAL NETWORK ALGORITHMS TO PREDICT RESTING ENERGY EXPENDITURE IN THE SAUDI ADULT POPULATION

PP413

ARTIFICIAL NEURAL NETWORK ALGORITHMS TO PREDICT RESTING ENERGY EXPENDITURE IN THE SAUDI ADULT POPULATION

A. M. Almajwal¹, Y. Almuhtadi¹, F. Mohammad², J. Al Muhtadi², M. M. A. Abulmeaty^1,*

¹Community Health Sciences, ²Center of Excellence in Information Assurance, King Saud University, Riyadh, Saudi Arabia

Rationale: Artificial neural networks can increase the predictability of equations, even for population-specific ones. This work aimed to improve the prediction of REE in the Saudi population using suitable AI tools.

Methods: The dataset of the previously published equation¹ was used to develop an artificial neural network (ANN) algorithm. Anthropometric and body composition parameters were used as proposed features.  The significant features were used to train the ANN model to accurately capture nonlinear correlations and make superior predictions. Subsequently, deep neural network (DNN) models, e.g., Extreme Gradient Boosting (XGBoost) & Convolutional Neural Network-Recurrent Neural Network (CNN-RNN), were used.

Results: A total of 423 participants (208 male) were divided into 3 sets: training (70%), validation (15%), and testing (15%). The ANN model and XGBoost develop two equations: AA_ANN1= 2.47 x BMI + 11.9 x AdjBW + 962.5 and AA_ANN2 = 4.29 x age + 9.4 x fat mass + 15.71 x FFMI + 1289.3, where BMI is Body Mass Index, AdjBW is Adjusted Body Weight (kg), and FFMI is Fat Free Mass Index (kg/m²). The AA_ANN1 presented a Root Mean Square Error (RMSE) of 215 and an accuracy of 66.2%, whereas AA_ANN2 presented a lower RMSE of 193 and a higher accuracy of 71.4%. The ANN model was trained on the top 10 features ranked by XGBoost, achieving an average accuracy of 90.2% (Table 1).

 Table 1: Comparison of predicted and measured REE.

Conclusion: The two new predictive equations, developed using an ANN combined with XGBoost, significantly improved REE prediction accuracy to 90.2%, thereby enabling its clinical application.

References: 1 Almajwal, A. M., & Abulmeaty, M. M. (2019). International journal of endocrinology, 2019(1); 5727496.

Disclosure of Interest: None declared