Artificial intelligence (AI) technology makes gait analysis based on two-dimensional images feasible. The first aim of this study is to develop an analysis platform that uses two-dimensional image analysis technology to determine the relative excursion of body segments and limbs in gait. The second aim is to apply this analysis platform to develop an artificial intelligence model for customized shoe adjustment to optimize gait. For the methodology, OpenPose will be used as the analysis tool to quantify the relative shifts between specific joint/node trajectories of the body, and the VICON three-dimensional motion analysis system will be used to verify the acquisition parameters. For the second aim, an existing 2D gait image database before customized shoe adjustment will be used to obtain the inputs to the customized shoe adjustment AI model. Through training the AI model, enabling to generate the outputs of the most appropriate adjustment parameters for shoe soles. After the AI model has well-trained, the walking kinematics and foot pressure before and after wearing ordinary sports shoes or customized shoes will be compared during running on a treadmill. This research is expected to establish an efficient customized shoe adjustment AI model, allowing users to improve walking efficiency.

Walking is a popular form of physical activity and exercise, but long-term walking is also associated with the risk of injury, so gait optimization should be considered. The application of biomechanics to injury prevention must meet practical requirements, including low cost, accessibility, rapid, effectiveness, and less physical effort. Footwear-based interventions have the opportunity to meet these requirements. Various designs of running shoes attempt to reduce injuries and improve performance, and adjustments to sole structure can have a significant impact on gait performance. Artificial intelligence (AI) technology makes gait analysis based on two-dimensional images feasible. The aim of this study for the first year is to develop an analysis platform that uses two-dimensional image analysis technology to determine the relative excursion of body segments and limbs in gait. The aim for the second year is to apply this analysis platform to develop an artificial intelligence model for customized shoe adjustment to optimize gait. Research methods. For the methodology, OpenPose will be used as the analysis tool in the first year to quantify the relative shifts between specific joint/node trajectories of the body, and the VICON three-dimensional motion analysis system will be used to verify the acquisition parameters. In the second year, an existing two-dimensional gait image database before customized shoe adjustment will be used to find the most relevant gait parameters using the aforementioned analysis platform, as the inputs to the customized shoe adjustment AI model. Through training the AI model, enabling to generate the outputs of the most appropriate adjustment parameters for customized shoe soles. After the AI model has well-trained, the walking kinematics and foot pressure before and after wearing ordinary sports shoes or customized shoes will be compared, as well as the differences in striking vibration and kinematics during running on a treadmill. This research is expected to establish an efficient customized shoe adjustment AI model, allowing users to achieve a smoother gait, more even foot pressure distribution, reduce joint loading, improve walking efficiency, and thereby reduce the risk of injury from long-term walking and running.

Inclusion Criteria:

• Self-percieved discomforts in lower extremity
• Be able to walk independently at least for 30 minutes and run independently at least for 10 minutes

Exclusion Criteria:

• Obvious foot deformity, trauma in the lower extremity
• Motor or neurologic disorder that affect walking patterns

Individuals with gait asymmetry or disorder