Study on Quantitative Evaluation Method of Balance Ability in Cancer Patients Based on Gait Features.
10.12455/j.issn.1671-7104.240656
- Author:
Junjie LIU
1
;
Xu ZHOU
2
;
Chao YU
2
;
Qingqing CAO
2
;
Zhiming YAO
2
;
Wanqiu ZHANG
3
;
Ling ZHANG
3
;
Wanqing YAO
3
;
Ning LIN
4
Author Information
1. School of Mechanical and Electrical Engineering, Anhui Jianzhu University, Hefei,
2. Institute of Intelligent Machines, Hefei Institutes of Physical Science, Chinese Academy of Sciences, Hefei,
3. Hefei Cancer Hospital, Chinese Academy of Sciences, Hefei,
4. Chuzhou Hospital Affiliated to Anhui Medical University (Chuzhou First People's Hospital), Chuzhou,
- Publication Type:Journal Article
- Keywords:
balance ability;
cancer patients;
gait analysis;
machine learning;
regression model
- MeSH:
Humans;
Postural Balance;
Neoplasms/rehabilitation*;
Gait;
Gait Analysis;
Biomechanical Phenomena;
Female
- From:
Chinese Journal of Medical Instrumentation
2025;49(4):369-374
- CountryChina
- Language:Chinese
-
Abstract:
The importance of gait assessment in the rehabilitation of cancer patients is gradually being recognized. However, quantitative analysis of balance ability in cancer patients is still limited. A total of 102 cancer patients meeting the inclusion criteria were recruited from Hefei Cancer Hospital, Chinese Academy of Sciences. Their balance ability was evaluated using the Berg Balance Scale (BBS). Gait data were collected by an electronic walkway and an IMU sensor system, including spatial-temporal and kinematic gait features such as step length, cadence, support time, and range of motion. Recursive feature elimination was used for feature selection. Ridge, Elastic Net, SVR, RF, and AdaBoost models were used to predict balance ability scores. Five-fold cross-validation was used to evaluate the performance of these models. Results show that the SVR model achieves the best performance with fifteen features (RMSE=3.22, R 2=0.91), followed by Ridge (RMSE=3.63, R 2=0.89). A method for evaluating balance ability based on gait features is proposed, providing a quantitative tool for personalized rehabilitation interventions in cancer patients.
