OC 05.1 - Machine learning models developed and internally validated for predicting chronicity in pediatric immune thrombocytopenia

Saturday, June 22, 2024

13:00 – 13:15 ICT

Room: 110 A-C

Presenting Author(s)

Jingyao Ma, MD

attending physician
Hematology Department, National Center for Children’s Health, Beijing Children’s Hospital, Capital Medical University, Beijing 101300, China.
Beijing, Beijing, China (People's Republic)

Co-Author(s)

cC

chang Cui

Ph.D.
State Key Laboratory of Multimodal Artificial Intelligence Systems, Institute of Automation, Chinese Academy of Sciences, Beijing, 100190, China
Beijing, Beijing, China (People's Republic)
YT

Yongqiang Tang

associate professor
State Key Laboratory of Multimodal Artificial Intelligence Systems, Institute of Automation, Chinese Academy of Sciences, Beijing, 100190, China
Beijing, Beijing, China (People's Republic)
YH

Yu Hu

Resident physician
Department II of Hematology Center, Beijing Children’s Hospital, Capital Medical University, National Center for Children’s Health, Beijing, 100045, China
Beijing, Beijing, China (People's Republic)
SD

Shuyue Dong

Resident physician
Department II of Hematology Center, Beijing Children’s Hospital, Capital Medical University, National Center for Children’s Health, Beijing, 100045, China
Beijing, Beijing, China (People's Republic)
Jialu Zhang

Resident physician
Department II of Hematology Center, Beijing Children’s Hospital, Capital Medical University, National Center for Children’s Health, Beijing, 100045, China
Beijing, Beijing, China (People's Republic)
XX

Xingjuan Xie

Resident physician
Department II of Hematology Center, Beijing Children’s Hospital, Capital Medical University, National Center for Children’s Health, Beijing, 100045, China
Beijing, Beijing, China (People's Republic)
JM

Jinxi Meng

Resident physician
Department II of Hematology Center, Beijing Children’s Hospital, Capital Medical University, National Center for Children’s Health, Beijing, 100045, China
Beijing, Beijing, China (People's Republic)
Zhifa Wang, PhD

Resident physician
Department II of Hematology Center, Beijing Children’s Hospital, Capital Medical University, National Center for Children’s Health, Beijing, 100045, China
Beijing, Beijing, China (People's Republic)
WZ

Wensheng Zhang

professor
State Key Laboratory of Multimodal Artificial Intelligence Systems, Institute of Automation, Chinese Academy of Sciences, Beijing, 100190, China
Beijing, Beijing, China (People's Republic)
Zhenping Chen

Associate professor
Hematologic Diseases Laboratory, Hematology Center, Beijing Pediatric Research Institute, Beijing Children's Hospital, Capital Medical University, National Center for Children’s Health, Beijing, 100045, ChinaChildren’s Health, Beijing, 100045, China
Beijing, Beijing, China (People's Republic)
Runhui Wu, PhD, MD

Professor
Department of Hematology, Beijing Children’s Hospital, Capital Medical University, National Center for Children’s Health, Beijing, 100045, China
Beijing, Beijing, China (People's Republic)

Background: Primary immune thrombocytopenia (ITP) in children is typically self-limiting; however, 20–30% of patients may experience prolonged thrombocytopenia lasting over a year. The challenge lies in predicting chronicity to ensure personalized treatment approaches.

Aims: To address this issue, we developed and internally validated four machine learning (ML) models that use demographic and immunological characteristics to predict the likelihood of chronicity.

Methods: The present study was conducted at Beijing Children's Hospital from June 2018 to December 2021, aiming to develop predictive models for determining the chronicity of pediatric ITP. Four ML models, based on a logistic regression classifier, a random forest classifier, eXtreme Gradient Boosting (XGBoost), and support vector machine, were employed. These models utilized a set of 16 variables including 14 immunological and 2 demographic predictors. The performance evaluation criteria included prediction accuracy, precision, recall, F1 score, and area under the ROC curve (AUC).

Results: Data were collected from 662 patients who were randomly assigned to either a training dataset or a testing dataset using a random number generator. Among them, 26.5% had chronic disease. All models performed well with AUC values ranging from 0.81 to 0.84, and XGBoost was selected for its highest AUC score and interpretability in constructing the predictive model. Age, Th17, Th17/Treg, TH1, and DNT were identified as significant predictors by the XGBoost algorithm.

Conclusion(s): We developed a precise predictive model for chronicity in pediatric ITP using ML during the initial phase. The XGBoost model achieved high predictive accuracy by utilizing individual patient clinical parameters and demonstrated commendable interpretability.