Title : Prediction of In-Hospital Cardiac Arrest Using Shallow and Deep Learning Journal : Diagnostics Authors : Minsu Chae, Sangwook Han, Hyowook Gil, Namjun Cho, Hwamin Lee Corresponding author : Hwamin Lee DOI : https://doi.org/10.3390/diagnostics11071255
Abstract Sudden cardiac arrest can leave serious brain damage or lead to death, so it is very important to predict before a cardiac arrest occurs. However, early warning score systems including the National Early Warning Score, are associated with low sensitivity and false positives. We applied shallow and deep learning to predict cardiac arrest to overcome these limitations. We evaluated the performance of the Synthetic Minority Oversampling Technique Ratio. We evaluated the performance using a Decision Tree, a Random Forest, Logistic Regression, Long Short-Term Memory model, Gated Recurrent Unit model, and LSTM–GRU hybrid models. Our proposed Logistic Regression demonstrated a higher positive predictive value and sensitivity than traditional early warning systems.
Acknowledgments This research was supported by the Bio and Medical Technology Development Program and Basic Science Research Program of the National Research Foundation (NRF) funded by the Korean government (MSIT) (No. NRF-2019M3E5D1A02069073 & NRF-2021R1A2C1009290) and Soonchunhyang University Research Fund.
Title : Prediction In-Hospital Cardiac Arrest within 8 hours using Vital signs Published in : The 16th Asia Pacific International Conference on Information Science and Technology (APIC-IST 2021) Author : Minsu Chae, Sangwook Han, Hyo-Wook Gil, Jun Ma, Min Hong, HwaMin Lee Corresponding author : HwaMin Lee Location : Paradise Hotel Busan, Korea
Abstract The rapid response system in the hospital uses the early warning score (EWS) to predict in-hospital cardiac arrest. However, the traditional EWS has low precision and low recall. Since the precision is less than 4%, there has a problem with false alarms. We performed a retrospective cohort study in Soonchunhyang University Cheonan Hospital, which is a tertiary teaching hospital in the Republic of Korea. We performed by changing the data slice size to 8, 16, 24, 32, 40, 48, 56 hours. The deep learning model implemented in this paper has higher precision and recall than the traditional EWS.
Acknowledgments This research was supported by Basic Science Research Program and the Bio & Medical Technology Development Program of the National Research Foundation (NRF) funded by the Korean government (MSIT) (NRF-2021R1A2C1009290 & No. NRF-2019M3E5D1A02069073). This research was supported by X-mind Corps program of National Research Foundation of Korea(NRF) funded by the Ministry of Science, ICT (No. 2019H1D8A1105622)
Title : A Comparative Study on the Effect of Eccentric Viewing Training Using PC and VR Contents Journal : International Journal on Advanced Science, Engineering and Information Technology Authors : Dokyeong Lee, HwaMin Lee Corresponding author : HwaMin Lee DOI : https://doi.org/10.18517/ijaseit.11.2.12832
Abstract The purpose of this study is to develop PC Eccentric Viewing Training (EVT) and VR EVT Content, to conduct experiments to verify the validity of contents first through non-disabled people, and to compare and analyze VR and PC contents. Both PC and VR contents were produced with UNITY. This content model assumes that reducing inaccurate saccades by eliminating eye movement helps improve reading accuracy. In addition, the two contents are implemented in the same way as VR and PC versions. The content consists of two steps, both PC and VR. The purpose of the content is to improve reading accuracy by improving the fixation stability of Preferred Retinal Locus (PRL) and reducing inaccurate Saccades. The experiment consisted of 12 persons (within maximum visual acuity less than 0.3), and they were assigned to the PC Content group and VR Content group. The experiment was conducted a total of 5 times, except for two weeks, which is the time to adapt PRL. The experimental results showed that the reading accuracy of the VR content group was higher. In addition, When comparing VR contents with PC contents, the group that conducted the training through PC contents showed a decrease in concentration as it progressed to 1-3 steps, and the score distribution also fell overall. In conclusion, the study compared VR and PC contents, and the effectiveness of contents was verified through experiments.
Acknowledgments This research was financially supported by the “ICT Convergence Smart Rehabilitation Industrial Education Program” through the Ministry of Trade, Industry & Energy (MOTIE). The authors are grateful to Korea Institute for Advancement of Technology (KIAT) and the MSIP (Ministry of Science, ICT and Future Planning), Korea, under the ITRC (Information Technology Research Center) support program (IITP-2020-2015-0-00403) supervised by the IITP (Institute for Information & communications Technology Promotion).
제목 : 머신러닝 기반의 약물 중독에 의한 호흡부전 예측 저자 : 한상욱, 채민수, 이화민 게재지 : 2021년 한국인터넷정보학회 춘계학술발표대회 장소 : 경주, 코모도호텔 경주 교신저자 : 이화민
초록 농촌지역 약물 중독으로 인한 사망자 수는 2020년 855명 이상으로 자궁암, 간염, 뇌종양, 식도암, 백혈병, 유방암, 알츠하이머 등의 질병으로 사망한 사람들의 숫자와 비교하여 결코 적지 않다. 우리나라 농업인의 급성 농약중독 유병률은 22.9~86.7%로 다양하다. 지역, 대상, 계절, 조사방법 등에 따라 상당한 차이가 있다. 농촌진흥청에서 정확한 피해 실태와 함께 그 전모를 파악하기 어려우며, 적절한 대책을 세울 수 없다고 설명하고 있다. 따라서 오직 예방 대책만이 강조되는 수준에 있다. 본 논문에서는 생체신호와 임상검사의 시계열 의료 데이터를 이용하여 약물 중독 환자의 호흡부전 증상을 예측하였다. 순천향대학교 천안병원에서 제공받은 1548명의 환자데이터를 기반으로 연구를 진행 하였으며, 생체정보와 Lab data를 합친 입력 요소 91개를 활용하였다.
Acknowledgments 이 논문은 2021년도 정부(과학기술정보통신부)의 재원으로 한국연구재단의 중견연구지원사업과 현장맞춤형 이공계 인재양성지원사업의 지원을 받아 수행된 연구임. (NRF-2021R1A2C1009290 & No. 2019H1D8A1105622).
제목 : 생징후 데이터를 이용한 8시간 이내 심정지 예측 저자 : 채민수, 한상욱, 한밀라, 홍민, 이화민 게재지 : 2021년 한국인터넷정보학회 춘계학술발표대회 장소 : 경주, 코모도호텔 경주 교신저자 : 이화민
초록 심정지는 심장이 멈추어 혈액이 흐르지 않는 현상이다. 혈액이 흐르지 않을 경우 신체 기능이 손상되며, 뇌손상 혹은 사망에 이를 수 있다. 그렇기 때문에 심정지 조기발견이 매우 중요하다. 심정지 조기 대응의 골든타임은 3분 이내 이다. 병원에서 신속대응팀에서 MEWS는 낮은 재현율과 높은 거짓알람의 문제점이 존재한다. 우리는 이러한 문제를 해결하기 위해 생징후 데이터를 사용하여 딥러닝 기반으로 8시간 이내 심정지를 예측하였다. 시퀀스 길이를 8시간, 16시간, 24시간, 32시간, 40시간, 48시간, 56시간, 64시간으로 나누어 실험하였다. 그 결과 시퀀스 길이가 56시간일 때 정밀도와 재현율이 가장 높았다. 시퀀스 길이를 56으로 고정하고 SMOTE 비율을 변경하여 0.1부터 1.0 까지 실험을 하였다. 그 결과 정밀도가 가장 높은 것은 SMOTE 비율을 0.04로 설정한 것이다. 시퀀스 길이가 56시간이고 SMOTE 비율이 0.04일 때 정밀도는 0.233, 재현율은 0.513, F1 Score는 0.301이다.
Acknowledgments 이 논문은 2021년도 정부(과학기술정보통신부)의 재원으로 한국연구재단의 중견연구지원사업과 바이오·의료기술개발사업의 지원을 받아 수행된 연구임. (NRF-2021R1A2C1009290 & No. NRF-2019M3E5D1A02069073).
Title : New Model for Predicting the Presence of Coronary Artery Calcification Journal : Journal of Clinical Medicine Authors : Samel Park, Min Hong, HwaMin Lee, Nam-jun Cho, Eun-Young Lee, Won-Young Lee, Eun-Jung Rhee, Hyo-Wook Gil Corresponding author : Eun-Jung Rhee, Hyo-Wook Gil DOI : https://doi.org/10.3390/jcm10030457
Abstract Coronary artery calcification (CAC) is a feature of coronary atherosclerosis and a well-known risk factor for cardiovascular disease (CVD). As the absence of CAC is associated with a lower incidence rate of CVD, measurement of a CAC score is helpful for risk stratification when the risk decision is uncertain. This was a retrospective study with an aim to build a model to predict the presence of CAC (i.e., CAC score = 0 or not) and evaluate the discrimination and calibration power of the model. Our data set was divided into two set (80% for training set and 20% for test set). Ten-fold cross-validation was applied with ten times of interaction in each fold. We built prediction models using logistic regression (LRM), classification and regression tree (CART), conditional inference tree (CIT), and random forest (RF). A total of 3302 patients from two cohorts (Soonchunhyang University Cheonan Hospital and Kangbuk Samsung Health Study) were enrolled. These patients’ ages were between 40 and 75 years. All models showed acceptable accuracies (LRM, 70.71%; CART, 71.32%; CIT, 71.32%; and RF, 71.02%). The decision tree model using CART and CIT showed a reasonable accuracy without complexity. It could be implemented in real-world practice.
Acknowledgments This research was supported by the Bio & Medical Technology Development Program of the National Research Foundation (NRF) funded by the Korean government (MSIT) (No. NRF-2019K1A3A1A20093097) and the Soonchunhyang University Research Fund.