Machine Learning Approach to Predicting Absence of Serious Bacterial Infection at PICU Admission.
Machine Learning Approach to Predicting Absence of Serious Bacterial Infection at PICU Admission.
复制标题
DOI：
10.1542/hpeds.2021-005998
复制
发表时间：
2022-06-01
影响因子：
--
通讯作者：
Bennett, Tellen D.
中科院分区：
文献类型：
--
作者：
Martin, Blake;DeWitt, Peter E.;Scott, Halden F.;Parker, Sarah;Bennett, Tellen D.
严重细菌感染(SBI)在PICU中很常见。抗生素可以减少相关的发病率和死亡率，但也有相关的不良影响。开发机器学习模型，能够识别SBI阴性儿童并减少不必要的抗生素。我们开发了使用生命体征、实验室和人口统计学变量来预测PICU入院时SBI阴性状态的模型。2011-2020年在我们的PICU住院的3个月至18岁的儿童，如果在24小时内进行感染评估，则包括在内，根据之前48小时内记录的抗生素暴露情况进行分层。受试者工作特征曲线下面积(AUROC)是主要的模型准确性衡量标准；其次，我们计算了每个模型确定为低风险的PICU中随后接受抗生素治疗的SBI阴性儿童的数量。15,074名儿童符合纳入标准。4788例(32%)在PICU入院前接受了抗生素治疗。在这些接触抗生素的患者中，2,325/4,788人(49%)有SBI。在10,286名未接触抗生素的患者中，2,356/10,286名(23%)有SBI。在抗生素暴露的儿童中，径向支持向量机模型在评估SBI方面具有最高的AUROC(0.80)，确定了48/442(11%)在PICU接受抗生素治疗的SBI阴性儿童，他们本可以为每个患者节省3.7天(IQR 0.9-9.0)抗生素的中位数。在未接触抗生素的儿童中，随机森林模型表现最好，但总体准确性较差(AUROC 0.76)，确定在PICU接受抗生素治疗的469名SBI阴性儿童中有33名(7%)，每个患者本可少用1.1天(IQR 0.9-3.7)抗生素。在PICU入院前接受抗生素治疗的儿童中，机器学习模型可以识别出低风险的SBI儿童，并有可能减少抗生素的暴露。
Serious bacterial infection (SBI) is common in the PICU. Antibiotics can mitigate associated morbidity and mortality but have associated adverse effects. To develop machine learning models able to identify SBI-negative children and reduce unnecessary antibiotics. We developed models to predict SBI-negative status at PICU admission using vital sign, laboratory, and demographic variables. Children 3-months-to-18-years-old admitted to our PICU, 2011–2020, were included if evaluated for infection within 24-hours, stratified by documented antibiotic exposure in the 48-hours prior. Area under the receiver operating characteristic curve (AUROC) was the primary model accuracy measure; secondarily, we calculated the number of SBI-negative children subsequently given antibiotics in the PICU identified as low-risk by each model. 15,074 children met inclusion criteria. 4,788 (32%) received antibiotics before PICU admission. Of these antibiotic-exposed patients, 2,325/4,788 (49%) had an SBI. Of the 10,286 antibiotic-unexposed patients, 2,356/10,286 (23%) had an SBI. In antibiotic-exposed children, a radial support vector machine model had the highest AUROC (0.80) for evaluating SBI, identifying 48/442 (11%) SBI-negative children given antibiotics in the PICU who could have been spared a median 3.7 (IQR 0.9–9.0) antibiotic-days per patient. In antibiotic-unexposed children, a random forest model performed best, but was less accurate overall (AUROC 0.76) identifying 33/469 (7%) SBI-negative children given antibiotics in the PICU who could have been spared 1.1 (IQR 0.9–3.7) antibiotic-days per patient. Among children who received antibiotics before PICU admission, machine learning models can identify children at low risk of SBI and potentially reduce antibiotic exposure.
登录
查看更多内容
影响因子：
3.2
作者：
Desautels T;Calvert J;Hoffman J;Jay M;Kerem Y;Shieh L;Shimabukuro D;Chettipally U;Feldman MD;Barton C;Wales DJ;Das R
通讯作者：
Das R
影响因子：
38.9
作者：
Fischer, JE;Ramser, M;Fanconi, S
通讯作者：
Fanconi, S
影响因子：
3.6
作者：
Gray, J;Gossain, S;Morris, K
通讯作者：
Morris, K
影响因子：
11.8
作者：
Donovan, Brittney M.;Abreo, Andrew;Wu, Pingsheng
通讯作者：
Wu, Pingsheng
影响因子：
4.9
作者：
Horan, Teresa C.;Andrus, Mary;Dudeck, Margaret A.
通讯作者：
Dudeck, Margaret A.