目的评价隐Markov模型在2型糖尿病不同发展阶段影响因素分析中的拟合稳健性。方法将697例2型糖尿病患者资料分成两个亚组,对2型糖尿病的资料用隐Markov模型进行重新拟合。结果两个亚样本模型参数相差基微。除职业因素外,有11项因素对2型糖尿病影响均有统计学意义(P<0.05);两个亚样本与原样本比较,其最大误差ME<0.05,两次隐Markov模型拟合结果一致。结论隐Markov模型拟合2型糖尿病的数据有较好的稳定性,可以用于慢性病生存质量影响因素资料的分析。 Objective To evaluate the robustness of Hidden Markov model in the influencing factors of different stages of type 2 diabetes mellitus. Methods The data of 697 patients with type 2 diabetes were divided into two subgroups. The data of type 2 diabetes were re-fitted with hidden Markov model. Results The two sub-sample model parameters differ slightly. In addition to occupational factors, there are 11 factors on the impact of type 2 diabetes were statistically significant (P <0.05); two sub-samples compared with the original sample, the maximum error ME <0.05, twice hidden Markov model fitted consistent results . Conclusion Hidden Markov model fitting data of type 2 diabetes has good stability and can be used to analyze the influencing factors of quality of life of chronic diseases.