Unsupervised phenotypic clustering for determining clinical status in children with cystic fibrosis
Unsupervised phenotypic clustering for determining clinical status in children with cystic fibrosis
复制标题
DOI：
10.1183/13993003.02881-2020
复制
发表时间：
2021-08-01
影响因子：
24.3
通讯作者：
Stanojevic, Sanja
中科院分区：
文献类型：
--
作者：
Filipow, Nicole;Davies, Gwyneth;Stanojevic, Sanja
背景 囊性纤维化（CF）是一种多系统疾病，仅根据肺功能评估疾病严重程度可能并不合适。该研究的目的是开发一种综合的机器学习算法来评估儿童独立于肺功能的临床状态。方法使用综合的前瞻性收集的临床数据库（加拿大多伦多）来应用无监督聚类分析。然后根据当前和未来的肺功能、未来住院的风险以及未来口服抗生素治疗的肺部病情恶化的风险对定义的集群进行比较。使用 k 最近邻 (KNN) 算法前瞻性地分配簇。这些方法在大奥蒙德街医院 (GOSH) 的儿科临床 CF 数据集中得到了验证。结果 最佳聚类模型根据 530 名个体的 12200 次接触识别出四个 (A-D) 表型聚类。与轻度疾病一致的两个簇（A 和 B）被确定为 1 秒用力呼气量（FEV1）较高，并且住院和口服抗生素治疗的肺部病情恶化的风险较低。还确定了与严重疾病一致的两个簇（C 和 D）具有低 FEV1。 D组的住院时间和口服抗生素治疗肺部病情恶化的时间最短。 GOSH 171 名儿童的 3124 次遭遇的结果是一致的。 KNN 聚类分配错误率较低，分别为 2.5%（多伦多）和 3.5%（GOSH）。 结论 机器学习衍生的表型聚类可以预测独立于肺功能的疾病严重程度，并且可以与功能测量结合使用来预测 CF 患者未来的疾病轨迹。
Background Cystic fibrosis (CF) is a multisystem disease in which the assessment of disease severity based on lung function alone may not be appropriate. The aim of the study was to develop a comprehensive machine-learning algorithm to assess clinical status independent of lung function in children.Methods A comprehensive prospectively collected clinical database (Toronto, Canada) was used to apply unsupervised cluster analysis. The defined clusters were then compared by current and future lung function, risk of future hospitalisation, and risk of future pulmonary exacerbation treated with oral antibiotics. A k-nearest-neighbours (KNN) algorithm was used to prospectively assign clusters. The methods were validated in a paediatric clinical CF dataset from Great Ormond Street Hospital (GOSH).Results The optimal cluster model identified four (A-D) phenotypic clusters based on 12200 encounters from 530 individuals. Two clusters (A and B) consistent with mild disease were identified with high forced expiratory volume in 1 s (FEV1), and low risk of both hospitalisation and pulmonary exacerbation treated with oral antibiotics. Two clusters (C and D) consistent with severe disease were also identified with low FEV1. Cluster D had the shortest time to both hospitalisation and pulmonary exacerbation treated with oral antibiotics. The outcomes were consistent in 3124 encounters from 171 children at GOSH. The KNN cluster allocation error rate was low, at 2.5% (Toronto) and 3.5% (GOSH).Conclusion Machine learning derived phenotypic clusters can predict disease severity independent of lung function and could be used in conjunction with functional measures to predict future disease trajectories in CF patients.