基于数据挖掘的艾滋病个性化治疗方案决策研究

发布时间：2018-04-03 01:39

本文选题：艾滋病　切入点：高效抗逆转录病毒治疗　出处：《首都医科大学》2017年硕士论文

【摘要】：目的:艾滋病是危害严重的一种传染性疾病。患者首次接受高效抗逆转录病毒治疗(highly active antiretroviral therapy,HAART)时的药物配伍方案是否合理是治疗成功与否的关键。由于艾滋病患者的个人身体素质不同、基线情况复杂、患病后病情进程不一、患者耐药性不同,选用某一药物配伍方案对这些患者进行治疗后,效果可能完全不同。本研究探讨利用数据挖掘方法分析艾滋病患者的多项指标与药物配伍方案之间的关系,制定艾滋病患者个性化治疗方案,从而避免治疗过程中的多次换药以及由此产生的耐药性、减轻药物治疗的毒副反应、提高患者生存质量、延长寿命。方法:收集721例首次接受HAART治疗的艾滋病患者治疗前以及治疗后多次随访时的年龄、身高、体重以及CD4+T淋巴细胞计数、CD8+T淋巴细胞计数、HIV病毒载量、白细胞计数、总淋巴细胞计数、血小板计数、血红蛋白、血肌酐、血尿素氮、血糖、谷草转氨酶、谷丙转氨酶、总胆红素数据,并对各种治疗方案中抗病毒药物的使用情况进行整理。对治疗后免疫重建良好患者的数据,分别通过基于患者相似性的聚类分析、基于案例推理和关联规则挖掘方法,确定患者的最佳治疗方案。最后对由三种方法得出的个性化治疗方案的准确性进行验证和比较。结果:(1)计算基于欧氏距离的患者相似性。利用聚类分析方法确定每一聚类中多数患者的用药方案为该类患者的最佳用药方案,由此推荐的患者用药方案与实际患者用药方案的一致率为67.3%。利用基于案例推理方法确定与某患者最相似的前10%患者的治疗方案作为该患者的推荐方案,分别以全部16项指标和7项核心指标计算相似性时,推荐方案与实际患者治疗方案的一致率分别为83.3%和86.0%。(2)经关联规则挖掘分析,得到用药情况与患者7个核心指标的相关关系,以及药物配伍情况之间对应关联。用药方案中包含依非韦伦的患者的最高支持度为0.02835,均高于用药方案中包含奈韦拉平的患者的最高支持度0.00622和包含洛匹那韦利托那韦的患者的最高支持度0.00138。留一法验证后的推荐一致率为71.4%。结论:利用数据挖掘技术中的聚类分析、基于案例推理和关联规则分析方法,能够针对艾滋病患者的人口学和实验室指标进行个性化用药方案推荐,其中基于案例推理方法的推荐准确性最高。研究证实了利用数据挖掘方法制定艾滋病患者个性化治疗方案的可行性和有效性,为患者个性化治疗方案的制定提供客观、科学的建议。
[Abstract]:Objective: AIDS is a serious infectious disease hazards. For the first time in patients receiving highly active antiretroviral therapy (highly active antiretroviral therapy, HAART) when the drug compatibility program is reasonable is the key to the success of treatment of AIDS patients. Due to personal physical fitness, the baseline situation is complex, the prevalence of disease after process, patients the resistance of these patients, treatment selection of a drug formula, the effect may be completely different. This study explores the use of data mining methods to analyze the relationship between a number of indicators of AIDS patients and drug combinations for patients, develop individualized treatment programs for AIDS, so as to avoid the repeated dressing treatment process and resulting resistance, reduce the side-effects of the drug treatment, improve the quality of survival, prolong the life of patients. Methods: collecting 721 cases of the first. By AIDS patients treated with HAART before and after treatment follow-up times of age, height, weight and CD4+T lymphocyte count, CD8+T lymphocyte count and HIV viral load, leukocyte count, total lymphocyte count, platelet count, hemoglobin, serum creatinine, blood urea nitrogen, blood glucose, aspartate aminotransferase, alanine aminotransferase, the total bilirubin data, and use of antiviral drugs in the treatment of a variety of sorting out. After the treatment of immune reconstitution in patients with good data, through the analysis of clustering similar patients based on case based reasoning and association rules mining method based on the best treatment plan were determined. Finally, verify and compare the accuracy of the individualized treatment plan obtained by the three methods. Results: (1) calculate the Euclidean distance similarity. Based on patients by using cluster analysis method to identify each cluster in the majority The medication regimen in patients with the best regimen for the patients, consistent rate of patients with medication and thus recommend medication for patients with the actual treatment of patients before 10% patients with the most similar as recommended the patient's case to determine the case based reasoning method using 67.3%., respectively, with all 16 indicators and 7 the core index of similarity calculation, consistency rate recommendation and treatment of actual patients were 83.3% and 86.0%. (2) by association rules analysis, get the correlation between the 7 core indicators and medication of patients, as well as the corresponding relationship between drug compatibility. Medication regimen included in the highest support of efavirenz with the degree of 0.02835, were higher than the highest support patients included nevirapine regimen in the highest support 0.00622 and contains the Tonave Wiley Los horse with a degree of 0.00138. After the verification of the method recommended the concordance rate was 71.4%. conclusion: using the technology of data mining and clustering analysis method, case based reasoning and association rules based on demographic and laboratory indicators for AIDS patients to recommend personalized medicine program, which recommended accuracy of case-based reasoning method based on maximum. Research has confirmed the use of data mining methods to develop the feasibility patients with personalized treatment of AIDS and effectiveness for patients with individualized treatment plan to provide objective and scientific advice.

【学位授予单位】：首都医科大学
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：R512.91;TP311.13

【参考文献】