基于随机空腹血糖（FPG）、糖化血红蛋白（HbA1c）及糖化血清蛋白（GSP）及聚类分析构建门诊精神分裂症（schizophrenia，SP）患者的分类判别预测模型，为SP患者共患糖代谢异常的早期诊治、监测管理和成本控制等策略的优化提供参考。

回顾2013年12月01日至2020年05月31于广州医科大学附属脑科医院门诊就诊SP患者的性别、年龄、门诊诊断、FPG、HbA1c及GSP等检验结果等资料，进行聚类及判别分析，并将分类后的数据进行对比分析。

（1）共纳入2047例患者的资料，门诊SP中年患者的各类血糖指标异常的比例较高，HbA1c（5.898±1.354），GSP（1.877±1.354），FPG（7.055±430）；（2）短期血糖指标监测中，男性升高的比例更高，差异具有统计学意义（P<0.05）；（3）可将其聚类分为3类，构建判别分类预测模型，第一类的模型为Y₁=-24.477+4.496HbA1c+6.781 GSP+1.641FPG，第二类的模型为Y₂=-139.639+6.404HbA1c+8.733GSP+8.592FPG，第三类的模型为Y₃=-49.354+5.502HbA1c+6.747GSP+3.831FPG；（4）对基于聚类的三组数据进行组间对比，发现这三类在年龄及HbA1c、GSP、FPG均存在显著差异（P<0.05）。

门诊SP患者血糖管理普遍较差，尤其是在男性的短期血糖指标管理上；基于血糖指标及聚类法构建的分类判别预测模型效果较好，可利用该结果探讨对门诊SP中年患者实施个体化监测，为此类糖代谢异常患者提供早期诊治。

Based on random fasting plasma glucose (FPG), glycosylated hemoglobin A1c (HbA1c) and glycosylated serum protein (GSP) and cluster analysis to construct a classification discriminant prediction model for middle-aged outpatients with schizophrenia (SP), which could provide reference for the early diagnosis and treatment, monitoring management and cost control strategies of SP patients with comorbidity of glucose metabolism.

from December 1, 2013 to May 31, 2020, the data of gender, age, outpatient diagnosis, FPG, HbA1c and GSP test results of SP patients in the outpatient department of Brain Hospital Affiliated to Guangzhou Medical University were reviewed, and the clustering and discriminant analysis were conducted, and the classified data were compared and analyzed.

(1) A total of 2047 patients were included in the study. The proportion of abnormal blood glucose indexes was higher in middle-aged patients with SP, HbA1c (5.898 ± 1.354), GSP (1.877 ± 1.354), FPG (7.055 ± 430); (2) In the short-term blood glucose monitoring, the proportion of male increased was higher, the difference was statistically significant (P<0.05); (3) The cluster can be divided into three categories, and the discriminant classification prediction model is constructed, the first kind is Y₁=-24.477+4.496HbA1c+6.781GSP+1.641FPG, the second kind is Y₂=-139.639+6.404HbA1c+8.733GSP+8.592FPG, the third one is Y₃=-49.354+5.502HbA1c+6.747GSP+3.831FPG; (4) Comparing the three groups of data based on clustering, it was found that there were significant differences in age, HbA1c, GSP and Glu among the three groups (P<0.05).

The results show that the management of blood glucose in outpatients with SP is generally poor, especially in the management of short-term blood glucose indicators of men; the classification discrimination prediction model based on blood glucose index and clustering method has a good effect, which can be used to explore the implementation of individualized monitoring for middle-aged patients with SP in outpatient department, so as to provide early diagnosis and treatment for patients with abnormal glucose metabolism.