To address the computational constraints of parameter estimation in the polytomous Cognitive Diagnosis Model (pCDM) in large-scale high data volume situations, this study proposes two two-stage polytomous attribute estimation methods: P_max and P_linear. The effects of the two-stage methods were studied via a Monte Carlo simulation study, and the applicability of the new methods was verified using an empirical data set from a national medical practitioners qualifying examination. The research found that in the face of massive data from large-scale assessments, the two-stage method can be used as an effective alternative to pCDM for assessing examinees' mastery patterns of polytomous attributes.