论文部分内容阅读
目的探讨部分可观察马尔可夫决策过程(POMDP)在住院2型糖尿病治疗方案优化研究中的应用问题。方法回顾性分析某院2012~2016年住院治疗的650例2型糖尿病患者的相关生化指标及临床用药数据等资料。采用MATLAB R2014a进行POMDP挖掘分析,以临床最大效应值的水平为标准,评价不同治疗方案的临床疗效。结果采用Gephi0.8.2筛选出治疗方案核心用药。POMDP模型评价了2型糖尿病临床治疗方案,发现优化的核心治疗方案。结论 POMDP模型可以从临床数据中挖掘出优化的治疗方案。
Objective To investigate the application of partially observable Markov decision making (POMDP) in the optimization of treatment plans for type 2 diabetes inpatients. Methods A retrospective analysis of a hospital from 2012 to 2016 hospitalized 650 cases of type 2 diabetes patients with biochemical indicators and clinical data and other data. Using MATLAB R2014a for POMDP mining analysis, the clinical efficacy of different treatment regimens was evaluated based on the level of clinical maximum effect value. Results Gephi0.8.2 screening out the core of the treatment program. The POMDP model evaluated the clinical management of type 2 diabetes and found an optimized core regimen. Conclusion The POMDP model can mine the optimal treatment plan from the clinical data.