Existing simplified calculation methods for critical load of pile piers in bridge structures exhibit significant errors in most cases， limiting their applicability. Based on calibration using field load tests on pile foundations in soft soil regions， a finite element method based on eigenvalue buckling analysis was employed to calculate the critical load of typical pile piers and compared with the results obtained from existing simplified calculation methods. The study identified the applicability range of the existing simplified calculation methods and found that they are not suitable for calculating pile piers in soft soil regions. Using a large dataset of finite element eigenvalue buckling analysis results for pile piers， the random forest algorithm was applied to analyze the calculation results， yielding a random forest model capable of calculating the length of pile piers and determining the importance index of various influencing parameters. Finally， based on the form of commonly used simplified calculation formulas and incorporating the important parameters identified by the random forest algorithm， a simplified calculation method for determining the critical load of pile piers in deeply soft soil regions was proposed through regression analysis.