Related Experiment Video
Updated: Jun 5, 2026
08:55
Use of a Foot-Induced Digitally Controlled Resistance Device for Functional Magnetic Resonance Imaging Evaluation in Patients with Foot Paresis
Published on: July 7, 2023
Meta-heuristic and machine learning based functional capacity prediction using gait parameters in patients with heart
Aylin Tanriverdi Eyolcu1, Ayşe Doğru2, Selim Buyrukoğlu3
1Department of Physiotherapy and Rehabilitation, Faculty of Health Science, Çankırı Karatekin University, Çankırı, Turkey. aylintanriverdi@karatekin.edu.tr.
Scientific Reports
|June 3, 2026
Summary
Wearable sensors and machine learning accurately predict heart failure (HF) patients' functional capacity using gait analysis. The PSO-CatBoost model, utilizing right stride length and gait speed, shows promise for clinical decision support.
Area of Science:
- Biomedical Engineering
- Cardiology
- Data Science
Background:
- Heart failure (HF) significantly impacts patients' functional capacity.
- Accurate prediction of functional capacity is crucial for HF management.
- Traditional assessment methods may have limitations.
Purpose of the Study:
- To predict functional capacity in HF patients using wearable sensor-based gait parameters.
- To evaluate meta-heuristic-based machine learning models for this prediction task.
- To identify key gait features influencing functional capacity.
Main Methods:
- Cross-sectional study with 70 HF patients.
- Gait parameters measured using wearable inertial sensors during the six-minute walk test.
- Machine learning models (XGBoost, LightGBM, Random Forest, CatBoost) optimized with meta-heuristics (Simulated Annealing, Genetic Algorithms, Particle Swarm Optimization, Bayesian Optimization).
- Synthetic Minority Over-Sampling Technique for Regression (SMOTeR) applied.
- 10-fold cross-validation used for performance evaluation (R², RMSE, MAE, MSE).
- SHapley Additive exPlanations (SHAP) used for model interpretability.
Main Results:
- The hybrid Particle Swarm Optimization-CatBoost (PSO-CatBoost) model achieved the highest predictive accuracy (R² = 0.9456).
- Key features identified by SHAP analysis were right stride length and gait speed.
- The PSO-CatBoost model demonstrated superior performance compared to other configurations.
Conclusions:
- The hybrid PSO-CatBoost model accurately predicts functional capacity in HF patients using gait parameters.
- Gait speed and stride length are significant predictors of functional capacity in this population.
- Further validation is needed to establish the clinical utility of this approach as a decision-support tool.
