Related Experiment Video
Updated: Apr 9, 2026
Virtual Agent for Real-Time Motivational Interviewing by Integrating Adaptive Nonverbal Behavior and Language Models
Published on: December 23, 2025
Artificial neural network modeling using clinical and knowledge independent variables predicts salt intake reduction
Hussain A Isma'eel1, George E Sakr1, Mohamad M Almedawar1
11 Division of Cardiology, Department of Internal Medicine, American University of Beirut, Beirut, Lebanon ; 2 Vascular Medicine Program, American University of Beirut Medical Center, Beirut, Lebanon ; 3 Department of Cardiovascular Medicine, Cleveland Clinic Foundation, Cleveland, Ohio, USA ; 4 Department of Electrical & Computer Engineering, 5 Department of Nutrition & Food Sciences, American University of Beirut, Beirut, Lebanon.
An artificial neural network (ANN) tool accurately predicts salt reduction behaviors (66%) in high cardiovascular risk patients. This model, outperforming traditional methods, can guide interventions for hypertension and cardiovascular disease prevention.
Area of Science:
- Cardiovascular Health
- Behavioral Science
- Artificial Intelligence in Medicine
Background:
- High dietary salt intake is a major risk factor for hypertension and cardiovascular diseases (CVDs).
- Accurate prediction of salt intake behaviors is crucial for effective intervention strategies.
- This study compares an artificial neural network (ANN) tool with least squares models (LSM) for predicting salt reduction behaviors.
Purpose of the Study:
- To compare the predictive accuracy of an ANN-based tool versus LSM for salt reduction behaviors.
- To evaluate the tool's performance using clinical data and knowledge questions in a high cardiovascular risk cohort.
- To assess the potential clinical utility of an ANN model for guiding therapeutic interventions.
Main Methods:
- Collected knowledge, attitude, and behavior data from 115 patients.
- Developed a reduced model with eight key knowledge items from an initial 69-item questionnaire.
- Calculated prediction accuracy using the bootstrap technique with 200 iterations for ANN and LSM.
Main Results:
- The ANN model achieved the highest prediction accuracy of 66% in the full model and 62% in the reduced model.
- Least Squares Models (LSM) showed significantly lower accuracy (40% full, 34% reduced).
- The ANN model demonstrated an average relative increase in accuracy of 82% (full) and 102% (reduced) over LSM.
Conclusions:
- Artificial neural network (ANN) modeling can predict salt reduction behaviors with 66% accuracy.
- The developed statistical model is implemented in an online calculator for clinical use.
- This tool can aid in tailoring therapeutic salt reduction interventions for individuals at high cardiovascular risk.
