Abstract: Prosthesis users are easier to fall down at the early stage. In order to solve this problem, a stumble pre-warning method based on the correlation analysis with fuzzy adaptive feedback regulation is proposed. Support vector machine(SVM) regression method is applied to improving correlation analysis template of different subjects, and the predictive value of two sensors are fused to distinguish stumble. For different patients, the body reaction is different in the case of stumble. A combined forecast algorithm based on fuzzy adaptive variable weight with feedback regulation unit is applied to parameter optimization for the weights of different patients, and the flexibility of stumble pre-warning system for lower limb prosthesis is improved. Finally, the feasibility of the proposed method is demonstrated through analyzing the data collected from several subjects. The experimental results show that the pre-warning correct rate is 95%. The system can be applied to rehabilitation training of patients in the early stage, so as to improve the safety of prosthesis.