로고

로고

Research

연구 정보

Research Project (승인과제목록)

KNN 연구 요약서

Title AI-driven respiratory distress syndrome prediction for newborn infants weighing less than 1500 g
Author 정성훈, 김지유, 최용성
작성자 최용성
Background In Korea, there are optional considerations on whether to administer surfactants to preterm infants regardless of whether they have respiratory distress syndrome (RDS). In particular, prophylactic surfactant administration is possible for preterm infants born less than 30 weeks of gestational age or less than 1,250 g of birth weight. However, not all preterm infants have RDS, so it is required to avoid unnecessary surfactant administration and positive pressure ventilation in preterm infants without RDS.
Aim / Hypothesis To develop a tool for predicting RDS by applying machine learning using pre- and peri-natal information as input data in very low birth weight infants (VLBWIs).
Inclusion Criteria All VLBWIs registered in KNN from 2013 Jan to 2020 Dec.
Exclusion Criteria 1) Missing data of parameter of “RDS” or “surfactant administration” 2) Congenital infection 3) Congenital anomalies
Study Design Statistical methods We will split the data into training and testing data at a ratio of 8:2 in a stratified fashion. The testing set will be used only for an independent test of our developed AI model. Using the training data, we will perform a 5-fold cross validation to confirm the generalization ability of the model. The training dataset was first randomly shuffled and divided into five equal groups in a stratified manner. Subsequently, four groups were selected for training the model, and the remaining group was used for internal validation. This process was repeated five times by shifting the internal validation group. Based on the 5-fold cross validation, we finalized our AI model, which is described in the following sections. We evaluated the performance of our AI model using the isolated testing data. AI Prediction Model Architecture We implemented and trained the proposed model using Tensorflow (version: tensorflow-gpu 2.6.0). We used Python (version: 3.7.4), Numpy (version: 1.21.5), Matplotlib (version: 3.5.1) and Scikit-learn (version: 1.0.2) to build the model and analyze the results.
Primary Outcomes RDS or surfactant administration
Secondary Outcomes and Definitions 1) 퇴원 전 사망 2) 입원 기간 중 패혈증 3) 입원 기간 중 ROP 4) 입원 기간 중 NEC 5) 입원 기간 중 IVH 6) 입원 기간 중 PVL 7) 입원 기간 중 수두증 8) 입원 기간 중 100ml/kg/day 장관 영양 도달 일자 9) 입원 기간 중 저혈압 약물 사용
Protocols 1) Eligibility criteria and demographics 2) 임신, 분만, 신생아 정보 3) 질환정보 I – 공기누출증후군, RDS, 폐표면활성제 사용여부, BPD예방을 위한 스테로이드사용 여부, 교정연령 36주 시점 및 기관지폐이형성증, 4) 질환정보 II – PDA 약물치료, 동맥관결찰술: 뇌실내출혈 단계, 뇌실주위 백질연화증, 패혈증 여부 있음/없음/미시행 5) 질환정보 III – 정맥영양기간, 괴사성장염 stage2 이상, 최초진단일, 괴사성 장염으로 인한 수술 여부, 미숙아 망막증 병기, 미숙아망막증 수술, 심각한 선천성 기형 여부, 선천성기형종류 6) 퇴원 관련 정보 – 퇴원일/입원일/재원일수, 퇴원형태 (가정/전원/사망/365이상입원중), 퇴원시 체중/신장/두위, 사망원인
Funding None

질병관리청 바로가기

대한신생아학회 바로가기

국립보건연구원 바로가기

Korean Neonatal Network TEL : 070-8095-2023   FAX : 02-3410-3540   
E-mail : knn@knn.or.kr

COPYRIGHT © 2022 KNN. ALL RIGHTS RESERVED.