With the deepening of population aging

fall prevention has become a critical measure to ensure the health and safety of the elderly.This paper elaborates on the specific workflow for constructing fall risk prediction models.By analyzing data collection methods

sensor types

data annotation processes

and model construction algorithms

this paper systematically reviews existing research findings.Particularly

it focuses on the new trends of multimodal data fusion and explainable artificial intelligence

while analyzing the limitations in the current fall risk prediction system.In the future

with the development of new sensor technologies and the standardization of data collection processes

fall prevention and control based on data fusion will have broad application prospects.