De-identified clinical dataset for: Identification of Clinical Phenotypes and Prediction Model for High-Risk Phenotypes of Pediatric Community-Acquired Pneumonia Based on Unsupervised Machine Learning

This dataset contains the de-identified clinical data used in the manuscript "Identification of Clinical Phenotypes and Prediction Model for High-Risk Phenotypes of Pediatric Community-Acquired Pneumonia Based on Unsupervised Machine Learning".Background: The study aimed to identify distinct clinical phenotypes in pediatric community-acquired pneumonia (CAP) patients and develop a prediction model for high-risk phenotypes using unsupervised machine learning.Data Source: The data was retrospectively collected from electronic medical records of pediatric CAP patients who underwent bronchoalveolar lavage at two hospitals in Fujian Province, China.Data Content: The dataset includes de-identified patient-level data on:- Demographic characteristics (e.g., age, gender)- Pathogen detection results (from BALF NGS, throat swab PCR, serology)- Inflammatory markers (WBC, NLR, CRP, PCT, LDH, D-dimer)- Imaging features- Clinical outcomes (length of hospital stay)- The assigned clinical phenotypes (MP-Dominant, Mixed-Infection, High-Inflammation) derived from k-prototypes clustering analysis.Usage Notes: This dataset is sufficient to reproduce the clustering analysis and predictive modeling results reported in the associated manuscript. All personally identifiable information has been removed to protect patient privacy.本数据集包含用于发表于论文《基于无监督机器学习的儿童社区获得性肺炎临床表型识别与高危表型预测模型》的去标识化临床数据。研究背景：本研究旨在识别儿童社区获得性肺炎（Pediatric Community-Acquired Pneumonia, CAP）患者的特异性临床表型，并通过无监督机器学习构建高危表型预测模型。数据来源：本数据集回顾性采集自中国福建省两家医院中接受支气管肺泡灌洗的儿童社区获得性肺炎患者的电子病历。数据内容：数据集包含去标识化的患者级数据，具体包括：- 人口统计学特征（如年龄、性别）- 病原体检测结果（源自支气管肺泡灌洗液（Bronchoalveolar Lavage Fluid, BALF）二代测序（Next-Generation Sequencing, NGS）、咽拭子聚合酶链反应（Polymerase Chain Reaction, PCR）及血清学检测）- 炎症标志物（白细胞计数（White Blood Cell, WBC）、中性粒细胞与淋巴细胞比值（Neutrophil-to-Lymphocyte Ratio, NLR）、C反应蛋白（C-Reactive Protein, CRP）、降钙素原（Procalcitonin, PCT）、乳酸脱氢酶（Lactate Dehydrogenase, LDH）、D-二聚体）- 影像学特征- 临床转归（住院时长）- 通过k-prototypes聚类分析得到的分配临床表型（支原体主导型、混合感染型、高炎症型）。使用说明：本数据集足以复现关联论文中报道的聚类分析与预测建模结果。所有个人可识别信息均已移除，以保护患者隐私。