An Improved Ngo-Grnn Algorithms for Accurately Predict Walking Energy Expenditure of Post-Stroke Patients by Easy Inputs

Abstract

Background and ObjectiveThe monitoring of walking energy expenditure (EE) plays an important role in functional evaluation and rehabilitation programs management for stroke patients. Traditional predictive equations developed solely based on walking speed showed significant heterogeneity, and are unable to accurately estimate walking EE of stroke patients.MethodsIn this paper, we propose an improved Northern Goshawk Optimization - Generalized Regression Neural Network (IMNGO-GRNN) for predict walking EE of stroke patients. Among of IMNGO, firstly, introduce the SPM chaotic mapping algorithm to the population initialization for increase population diversity; secondly, incorporate Sine Cosine Algorithm to the exploration phase of original NGO for improve global optimization ability. The leave one out cross validation, ablation and comparative studies were conducted to examine the performance of the proposed model for predicting walking EE of stroke patients.ResultsThe significantly strong correlation were found between the predicted and measured EE in the training and testing data (R2 = 0.9864; 0.9483); among of five model algorithm variants, the IMNGO-GRNN with the highest prediction accuracy; compared to previous study, the proposed model reduced the estimation error by 86.49%.ConclusionsThe proposed IMNGO-GRNN algorithm model that can achieve the high-precision prediction of treadmill walking EE for stroke patients based on the easy-obtain inputs. This algorithm model offered smarter solution for convenient monitoring walking EE of stroke patients does not depend on complex measurement systems and signals, and have potential application value in health science and rehabilitation medicine.

Note:
Funding declaration: This study was supported by National Key Research and Development Program of China (2023YFC3603700), National Natural Science Foundation of China (12272273), Science and Technology Development Fund of Shanghai Pudong New Area (KJW2024-Y18), Medical Engineering Cross Innovation Special Project of Shanghai Seventh People's Hospital (QYYGZ0202), Traditional Chinese Medicine Peak Discipline Construction Project of Pudong New Area Health Commission (YC-2023-0601), Shiyinyu National Famous Traditional Chinese Medicine Inheritance Studio (YC-2023-0120) and Key Department Construction Project of Collaborative Traditional Chinese and Western Medicine.

Conflict of Interests: None of the authors have a conflict of interest to disclose.

Ethical Approval: The study was approved by the Institutional Review Board (SBKT-2022-010) and informed consent was obtained from all participants.

Keywords: energy expenditure, stroke, walking, northern goshawk optimization, generalized regression neural network