Department of Emergency, Kunming Yan'an Hospital, Kunming 650000, Yunnan, China.
Comput Intell Neurosci. 2022 Mar 7;2022:3387212. doi: 10.1155/2022/3387212. eCollection 2022.
Adaptive niche genetic algorithm (ANGA) and lung ultrasound were combined, the death warning mathematical model was established for patients with sepsis-lung injury, and the epidemiological characteristics were analyzed to explore the efficacy of Vancomycin in the treatment of sepsis-lung injury. First, 88 sepsis patients with lung injury were selected as the research objects. General clinical data and pulmonary ultrasound results were collected. On this basis, epidemiological analysis was carried out, and the death warning model of patients with sepsis-lung injury was established based on ANGA algorithm. Then, the total cure rate, (SA) clearance rate, methicillin-resistant SA (MRSA) clearance rate, and the incidence of adverse reactions after intravenous infusion of Vancomycin were analyzed. The results showed that the ANGA mathematical model combined with the random forest (RF) classifier proposed had better classification effect and robustness relative to the traditional principal component analysis and NGA. The early warning accuracy of the proposed ANGA + RF mathematical model was higher than 95% in contrast to that of the APACHE-II score and the SOFA score. Compared with patients in the severe group, the MRSA infection rate and the levels of procalcitonin (PCT), C-reactive protein (CRP), and activated partial thromboplastin time (APTT) of SA sepsis-lung injury patients were greatly reduced, while thrombin time (TT) and D-D dimer in the death group were considerably increased ( < 0.05), and the PLT level was greatly reduced ( < 0.05). In addition, the total cure rate, SA clearance rate, and MRSA clearance rate of Vancomycin-treated SA sepsis-lung injury patients were significantly increased ( < 0.05) compared with patients in the conventional treatment control group. However, the probability of adverse reactions was increased notably ( < 0.05). ANGA combined with RF classifier can improve the accuracy of death warning in patients with sepsis-lung injury. Vancomycin can effectively eliminate MRSA infection rate in patients with sepsis-lung injury and improve the treatment effect of patients.
自适应生态位遗传算法（ANGA）与肺部超声相结合，建立脓毒症肺损伤患者死亡预警数学模型，并进行流行病学特征分析，探讨万古霉素治疗脓毒症肺损伤的疗效。首先，选取 88 例肺部损伤的脓毒症患者作为研究对象，采集其一般临床资料和肺部超声结果。在此基础上进行流行病学分析，基于 ANGA 算法建立脓毒症肺损伤患者死亡预警模型。然后，分析患者的总治愈率、（SA）清除率、耐甲氧西林 SA（MRSA）清除率及静脉滴注万古霉素后的不良反应发生率。结果表明，与传统主成分分析和 NGA 相比，基于随机森林（RF）分类器提出的 ANGA 数学模型具有更好的分类效果和鲁棒性。与 APACHE-II 评分和 SOFA 评分相比，所提出的 ANGA+RF 数学模型的早期预警准确率均高于 95%。与重症组患者比较，SA 脓毒症肺损伤患者中 MRSA 感染率以及降钙素原（PCT）、C 反应蛋白（CRP）、活化部分凝血活酶时间（APTT）水平明显降低，而死亡组患者的凝血酶时间（TT）和 D-二聚体明显升高（<0.05），血小板计数（PLT）水平明显降低（<0.05）。此外，与常规治疗对照组患者比较，万古霉素治疗 SA 脓毒症肺损伤患者的总治愈率、SA 清除率和 MRSA 清除率均明显升高（<0.05），但不良反应发生率显著升高（<0.05）。ANGA 结合 RF 分类器可提高脓毒症肺损伤患者死亡预警的准确性，万古霉素能有效消除脓毒症肺损伤患者的 MRSA 感染率，提高患者的治疗效果。
