a deployable gait-based fall-risk surveillance tool for older adults using deep learning in public health and care settings
1
School of Public Health, Shantou University, Shantou, China
2
Department of Orthopedics, Guangdong Provincial People's Hospital (Guangdong Academy of Medical Sciences), Guangzhou, China
Popul. Med. 2026;8(Supplement Supplement 1):A43
ABSTRACT
INTRODUCTION:
falls are a leading and largely preventable cause of morbidity disability and health care burden among older adults particularly in institutional and hospital settings.¹,² existing fall-risk screening tools often rely on subjective assessments and lack scalability for routine surveillance.³ objective gait-based assessment offers a non-invasive and reproducible approach to early risk identification.⁴,⁵ this study aimed to develop a compact and interpretable gait-based fall-risk prediction framework suitable for real-world clinical and care environments.
METHODS:
a total of 207 inpatients aged 65 to 93 years from a tertiary hospital were assessed using a bedside motion analysis system capturing six-degree-of-freedom knee ankle and hip kinematics during a 15 second walking trial. gait time-series data were processed using a lightweight residual one-dimensional convolutional neural network with dual attention mechanisms. the workflow integrated gait data acquisition preprocessing risk classification and interpretability modules into a unified bedside pipeline. fall-risk categories were defined as low moderate or high according to a validated clinical scale.⁶ model performance was evaluated using accuracy precision recall f1 score area under the receiver operating characteristic curve and confusion matrices.
RESULTS:
the proposed model achieved an overall accuracy of 0.878 with macro precision of 0.874 macro recall of 0.897 macro f1 of 0.882 and an area under the receiver operating characteristic curve of approximately 0.95. recall for the high-risk group reached 100 percent indicating effective identification of individuals at greatest risk. model interpretation highlighted specific gait phases associated with instability supporting transparent linkage between biomechanical features and risk classification.⁷,⁸
CONCLUSIONS:
this interpretable gait-based bedside workflow enables objective identification of fall risk among older adults using minimal space and automated processing. its compact and scalable design supports integration into routine clinical and care settings and may contribute to fall-risk surveillance and early prevention strategies in aging populations.¹,³,⁵
falls are a leading and largely preventable cause of morbidity disability and health care burden among older adults particularly in institutional and hospital settings.¹,² existing fall-risk screening tools often rely on subjective assessments and lack scalability for routine surveillance.³ objective gait-based assessment offers a non-invasive and reproducible approach to early risk identification.⁴,⁵ this study aimed to develop a compact and interpretable gait-based fall-risk prediction framework suitable for real-world clinical and care environments.
METHODS:
a total of 207 inpatients aged 65 to 93 years from a tertiary hospital were assessed using a bedside motion analysis system capturing six-degree-of-freedom knee ankle and hip kinematics during a 15 second walking trial. gait time-series data were processed using a lightweight residual one-dimensional convolutional neural network with dual attention mechanisms. the workflow integrated gait data acquisition preprocessing risk classification and interpretability modules into a unified bedside pipeline. fall-risk categories were defined as low moderate or high according to a validated clinical scale.⁶ model performance was evaluated using accuracy precision recall f1 score area under the receiver operating characteristic curve and confusion matrices.
RESULTS:
the proposed model achieved an overall accuracy of 0.878 with macro precision of 0.874 macro recall of 0.897 macro f1 of 0.882 and an area under the receiver operating characteristic curve of approximately 0.95. recall for the high-risk group reached 100 percent indicating effective identification of individuals at greatest risk. model interpretation highlighted specific gait phases associated with instability supporting transparent linkage between biomechanical features and risk classification.⁷,⁸
CONCLUSIONS:
this interpretable gait-based bedside workflow enables objective identification of fall risk among older adults using minimal space and automated processing. its compact and scalable design supports integration into routine clinical and care settings and may contribute to fall-risk surveillance and early prevention strategies in aging populations.¹,³,⁵
| eISSN: | 2654-1459 |
We process personal data collected when visiting the website. The function of obtaining information about users and their behavior is carried out by voluntarily entered information in forms and saving cookies in end devices. Data, including cookies, are used to provide services, improve the user experience and to analyze the traffic in accordance with the Privacy policy. Data are also collected and processed by Google Analytics tool (more).
You can change cookies settings in your browser. Restricted use of cookies in the browser configuration may affect some functionalities of the website.
You can change cookies settings in your browser. Restricted use of cookies in the browser configuration may affect some functionalities of the website.
