유럽동맥경화학회 예측모형을 이용한 한국인 심혈관 질환 사망 연구

지슬기 연세대학교 보건대학원 2015 국내석사

This study has a purpose to estimate Korean’s cardiovascular disease risk of mortality in ten years, by applying risk of mortality from cardiovascular disease estimation (SCORE), developed in Europe, into the study objects 3,609 adults from 40 to 79 years old by utilizing data in 2012 among 5th primitive data of Korea National Health and Nutrition Examination Survey. This study estimated risk of mortality in ten years by utilizing demographic factors as gender and age and clinical factors as smoking, systolic blood pressure, total cholesterol, and HDL cholesterol. This study with calculated estimation of risk of mortality from cardiovascular disease in ten years added and analyzed socio-economic factors for relevance analysis. Among the total 3,609 research objects, the number of male is 1,541, which are 42.7%, and the number of female is 2,068, which is 57.3%. Prediction scoring of Korean’s risk of mortality from cardiovascular disease in ten years is high both models of SCORE-High and SCORE-Low, when comparing risk charts suggested by European Society of Cardiology. The male objects receive higher scoring than the female objects, and as they are older their scoring is higher. Especially, in case of the male objects, some objects that are smoking have very high scoring of risk of mortality from cardiovascular disease. Also, in case of female objects from 40 to 54 years old, the scoring is comparatively low regardless of smoking, and in case of male objects over 55 years old risk of mortality from cardiovascular disease is high regardless of smoking. When SCORE-High model is applied, the men’s average scoring of risk of mortality from cardiovascular disease in ten years is 6.2% and women’s is 2.5%. When SCORE-Low model is applied, the men’s average scoring of risk of mortality from cardiovascular disease in ten years is 3.5% and the women’s is 1.7%. In case of male objects, these numerical values are three to six times higher and in case of female objects these numerical values are four to six times higher than numerical values suggested from the latest domestic research (Jee et al., 2014). Based on prediction scoring of risk of mortality from cardiovascular disease in ten years that Conroy et al. (2003) suggested, men’s the number of death toll from 40 to 79 year-old Korean is 503,814 and women’s is 227,477. If it is assumed that a similar number of deaths occur in each year, this study could estimate that the number of death toll from cardiovascular disease becomes 50,381 and 22,748 each of men and women. Indeed, assumed by the Europe’s SCORE model, the number of death toll from cardiovascular disease is 73,129 ( the number of men’s is 50,381 and the number of women’s is 22,748). The total number of deaths that domestic statistics reported was 267,221 (the number of men’s is 147,372 and the number of women’s is 119,849 in 2012. Among the number, the number of deaths from circulatory system disease was 18,531 in case of male and 12,017 in case of female. As the prediction number from the Europe’s SCORE model is divided into the actual number of deaths, the Europe’s model produces more excessive estimation that is 2.7 times in case of male and 1.9 times in case of female than the actual number of deaths of Korean. In each age, the biggest excessive estimation occurs in case of men as 3.9 times more from 50 to 54 years old and in case of women as 3.5 times more from 60 to 64 years old. In conclusion, because, based on SCORE estimation, the Korean’s prediction scoring of risk of mortality from cardiovascular disease is over estimated regardless of age and gender, it should be needed to develop modified SCORE estimation to be appropriate on the Korean situations. 이 연구의 목적은 국민건강영양조사 제5기 자료 중 2012년 자료를 이용하여 40세부터 79세 사이의 성인 남녀 3,609명의 대상자들에게 유럽에서 개발한 심혈관 질환 사망 위험도 추정식 (SCORE)을 적용하여 한국인의 심혈관 질환 10년 사망 위험도를 추정하는 것이다. 심혈관 질환 10년 사망 위험도 산정은 인구학적인 요인으로 성별과 연령, 임상적 요인으로 흡연 여부, 수축기 혈압, 총콜레스테롤, HDL 콜레스테롤 요인을 활용하였다. 심혈관 질환 10년 사망 위험도 산정 수치를 가지고 관련성 분석을 위해 사회경제학적인 요인을 추가하여 분석하였다. 연구 대상자 총 3,609명 중 남자는 1,541명으로 42.7%, 여자는 2,068명으로 57.3%이다. 한국인의 심혈관 질환 10년 사망 위험도 예측 점수는 유럽심장학회에서 제시하는 위험차트와 비교할 때 SCORE-High와 SCORE-Low 모형 모두에서 높았다. 여성 대상자보다는 남성 대상자가, 나이가 많을수록 높았다. 특히 남성 대상자 중에서 흡연자들의 경우 심혈관 질환 사망의 위험도 점수가 매우 높았다. 또한 40세부터 54세의 여성 대상자인 경우 심혈관 질환 사망 위험이 흡연 여부와 상관없이 비교적 낮으며, 55세 이상의 남성 대상자인 경우 흡연 여부에 상관없이 심혈관 질환 사망 위험도가 높았다. SCORE-High를 적용할 경우 심혈관 질환 10년 사망 위험의 평균 점수는 남성인 경우 6.2%, 여성인 경우 2.5%였다. SCORE-Low를 적용할 경우 심혈관 질환 10년 사망 위험의 평균 점수는 남성인 경우 3.5%, 여성인 경우 1.7%였다. 이것은 최근의 국내연구(Jee et al., 2014)에서 제시한 수치보다 남성은 3-6배, 여성은 4-6배 높았다. Conroy et al.(2003)에서 제시한 심혈관 질환 10년 사망 위험도 예측 점수를 기준으로 한국인의 40세부터 79세 사이의 10년간 평균 기대 심혈관 질환 사망수는 남자 503,814명, 여자 227,477명이었다. 이를 연간 비슷한 수가 사망한다고 가정하면 연간 심혈관 질환 사망수는 남녀 각각 50,381명, 22,748명으로 추정된다. 즉 유럽의 SCORE 모형에 의해 추정된 1년간 심혈관 질환 관련 사망수는 73,129명 (남자 50,381, 여자 22,748명)이었다. 2012년 우리나라 통계청에서 보고한 전체 사망자수는 267,221명 (남자 147,372명, 여자 119,849명)이었다. 이중에 40-79세 중 순환기 계통의 질환으로 사망한 수는 남자 18,531명, 여자 12,017명이었다. 이를 유럽 SCORE 모형에 의한 기대수를 실제 사망수로 나누면, 유럽 모형은 한국인의 실제 사망수보다 남자 2.7배, 여자 1.9배 과대 추계됨을 알 수 있다. 연령별로는 남자의 경우 50-54세 연령이 3.9배로 과대추계 규모가 가장 컸고, 여자는 60-64세가 3.5배로 과대 추계 규모가 가장 컸다. 결론적으로 SCORE 추정식에 의한 한국인의 심혈관 질환 사망의 위험도 예측 점수는 연령 및 성별에 관계없이 과대 추정되고 있으므로 한국인 실정에 맞는 보정된 SCORE 추정식을 개발해야 할 것이다.

A Study on the Machine Learning-Based Prediction Algorithm Models for Cardiovascular Disease : A Study on the Prediction Algorithm Models for Cardiovascular Disease

김경아 을지대학교 대학원 2025 국내박사

A Study on the Machine Learning-Based Prediction Algorithm Models for Cardiovascular Disease Kyung-A Kim Department of Medical Artificial Intelligence, Graduate School, Eulji University (Supervised by Professor Myung-Ae Chung, Ph.D.) The recent rapid development of artificial intelligence (AI) technology is leading to innovative changes in the medical field, and AI-based technologies are actively being utilized in the prediction and management of major diseases such as cardiovascular disease. Cardiovascular disease is a disease that ranks among the top causes of death worldwide, and early prediction and prevention of it is a very important task in public health. Along with the rapid growth of the digital healthcare market, the convergence of AI and medical devices is suggesting a new paradigm for the management and treatment of cardiovascular diseases. The purpose of this study is to develop a digital therapeutic (Software as a Medical Device, SaMD) to provide customized lifestyle improvement content to patients with cardiovascular disease, and to build a cardiovascular disease prediction model and compare and analyze its performance. To this end, various machine learning algorithms such as Logistic Regression (LR), Linear Discriminant Analysis (LDA), Support Vector Machine (SVM), and XGBoost were utilized, and the performance of each model was evaluated and compared. The data used in the study were based on domestic and international public data and clinical data, and included various variables that can affect cardiovascular disease, such as gender, age, blood pressure, blood sugar, and past disease history. In the data preprocessing process, data quality was improved through missing value processing and standardization, and appropriate data partitioning was performed for learning and evaluation of the machine learning model. Additionally, in order to select the optimal algorithm to be used as a prediction model, the time complexity and space complexity of the algorithms were calculated and compared. The model performance was evaluated using indicators such as accuracy, F1 Score, and ROC-AUC. As a result of the study, the XGBoost model showed superior prediction performance compared to other algorithms. The accuracy of the XGBoost algorithm was approximately 90%, the F1 Score was in the range of 0.80 to 0.82, and the ROC-AUC was 96%. Based on this performance, the XGBoost model was evaluated to be able to be used as a reference model for predicting cardiovascular disease. In particular, when data collected through clinical research were applied to the same XGBoost model, the prediction accuracy was over 90%, proving the high reliability of the model. Another important contribution of this study is the development of a mobile app and web-based digital therapeutics that provide customized content for improving lifestyle habits based on the research results. The app provides functions for predicting and managing the risk of cardiovascular disease in real time. In addition, the app performance evaluation results using clinical data and public data showed the same performance as the prediction accuracy of the XGBoost algorithm model. However, continuous advancement of cardiovascular prediction models and digital therapeutics is necessary, and additional clinical studies should be conducted to establish scientific and clinical grounds for digital therapeutics (software medical devices) that provide evidence-based therapeutic interventions to patients to prevent, manage, and treat medical disorders or diseases. Despite these limitations, the cardiovascular disease prediction model and app- and web-based digital therapeutics developed in this study are expected to provide practical help in the prevention and management of cardiovascular disease. In the future, by applying this digital therapeutic to clinical patients, it will be possible to increase the possibility of early prediction and treatment of cardiovascular diseases and contribute to health promotion through improvement of lifestyle habits. Furthermore, along with the development of the digital healthcare industry, the results of this study are expected to become a key element of future medicine. Keywords : Machine Learning, Cardiovascular Disease, Algorithm, XGBoost, Digital Healthcare, Digital Therapeutics

사업장 근로자의 생활습관 및 심혈관질환 인식과 심혈관질환 위험인자와의 관련성

김정희 가천의과학대학교 대학원 2011 국내석사

목적 : 급속한 산업화와 경제성장으로 질병의 양상도 변화되어 생활습관 관련 질환이 문제시 되고 있다. 이러한 잘못된 생활습관에 의해 1차적으로 발생하는 심혈관질환을 예방하기 위해 생활습관에 대한 지속적인 관리가 필요하다. 그러나 사업장 근로자들은 급격한 산업화에 따른 경제, 사회 및 생활습관의 변화에 따라 건강상 많은 어려움을 겪고 있다. 이에 본 연구는 사업장 근로자의 생활습관과 심혈관질환에 대한 인식을 파악하고 심혈관질환 위험인자와의 관련성을 확인함으로써 사업장 근로자의 건강증진 및 질병예방 전략을 수립하기 위한 기초자료를 제공하고자 한다. 대상 및 방법 : 연구대상은 2011년 3월 14일부터 5월 13일까지 인천시내에 소재한 사업장 근로자로써 N병원에 건강검진을 받으러 온 근로자 중 심혈관질환 기왕력자를 제외한 559명을 대상으로 설문조사를 실시하였고, 건강검진 검사결과 자료를 수집하였다. 자료분석은 SPSSWIN (ver.17.0)을 이용하여 실수와 백분율, 일표본 t-검정, Kruskal-Wallis와 Mann-Whitney U 검정법, Spearman의 순위상관 검정법으로 검정하였다. 결과 : 사업장 근로자들의 심혈관질환의 인식도는 높은 편이였다. 심혈관질환 위험인자에서 BMI, SBP와 DBP, FBS, TC, HDL-C, LDL-C가 대부분 정상이었다. 생활습관에 따른 심혈관질환 위험인자의 특성에서 흡연군보다 비흡연군이 SBP, DBP, FBS, TC가 낮았고, HDL-C는 높게 나타났으며, 고흡연군보다 저흡연군이 SBP와 DBP가 낮았다. 음주군보다 비음주군이 BMI, SBP, DBP, TC가 낮게 나타났으며, 규칙적인 운동 습관과 식습관을 가진 근로자들이 HDL-C가 높았다. 심혈관질환 인식과 위험인자 간에는 상관성이 없는 것으로 나타났다. 결론 : 사업장 근로자들의 심혈관질환 위험인자들은 대부분 정상이었고, 심혈관질환에 대한 인식은 높은 편이었으나, 생활습관 중 규칙적인 식습관을 제외한 흡연, 음주, 운동, 염분섭취량과 심혈관질환 인식 간에는 차이가 없는 것을 볼 때 근로자들이 올바른 생활습관의 필요성은 인지하고 있으면서도 실천을 못하고 있는 것으로 나타났다. 따라서 근로자들의 건강을 유지하고 증진을 도모하기 위해 직장 내 건강증진 프로그램 등이 지속적으로 운영되어야 하며, 근로자들이 건강한 생활습관을 실천할 수 있는 방안을 모색해야 할 것이다. Purpose : As the rapid industrialization and economic growth change the patterns of the disease, the diseases related with life style is also a concern. Ongoing management about life style is needed in order to prevent cardiovascular disease occurred primarily by these bad habits. However, industrial workers are suffering a lot of health difficulties due to economic, social and life style changes depending on the rapid industrialization. The purpose of this study is to provide a basic data to establish the industrial worker's health promotion and disease prevention strategy by checking the relevance of cardiovascular disease risk factors and identifying industrial worker's life style and cardiovascular disease awareness. Materials and Methods : The study target was industrial workers located within Incheon for a health checkup in the N hospital from 2011 March 14 to May 13 except for the history of cardiovascular disease and 559 people were surveyed by utilizing health screening tests and other materials. Data analysis was tested as numbers and percentages, single sample t-test, Kruskal-Wallis and Mann-Whitney U test, Spearman's rank correlation by utilizing SPSSWIN(ver.17.0). Results : The industrial worker's awareness of cardiovascular disease was high. BMI, SBP and DBP, FBS, TC, HDL-C, LDL-C in the risk factors for cardiovascular disease were mostly normal. As for the nature of the risk factors of cardiovascular disease according to life style, non-smoker group had lower SBP, DBP, FBS, TC and higher HDL-C than smoker group, light smoker group showed lower SBP and DBP than heavy smoker group. Non-drinking group had lower BMI, SBP, DBP, TC than drinking group and workers with regular exercise and eating habits showed higher level of HDL-C. There was no correlation between perception and risk factors for cardiovascular disease. Conclusion : The industrial worker's risk factors for cardiovascular disease were mostly normal and awareness for cardiovascular disease was high, but there was no difference between smoking, drinking, exercise, salt intake and cardiovascular disease awareness except for regular eating habits in life style. The following conclusion was that workers recognized the necessity of right life style but they couldn't practice it well. Therefore, health promotion programs in the workplace should be operated continuously to maintain and promote the worker's health and ways should be found to practice healthy life style for workers.

Developmental pathways from adverse childhood experiences to risk of cardiovascular disease : the role of substance use and allostatic load

허윤성 Graduate School, Yonsei University 2024 국내석사

본 연구는 아동기 역경 경험이 청소년 약물 사용과 알로스태틱 부하를 통해 성인기 심혈관 질환의 위험에 미치는 종단적 영향을 분석하였다. 이를 위해 미국의 대표적인 종단 패널인 Add Health의 1차(아동기) ~ 4차(성인기)의 자료가 사용되었으며, 주요 변수에 대한 응답값을 모두 보유하고 있는 3,254명이 분석에 포함되었다. 최종 연구 모형은 Mplus 8.7을 통해 구조 방정식 모형을 분석하였다. 본 연구의 주요 결과는 다음과 같다. 첫째, 아동기 역경 경험, 청소년기 약물 사용, 알로스태틱 부하는 성인기 심혈관 질환 위험에 직접적인 정적 영향을 미치는 것으로 나타났다. 둘째, 아동기 역경 경험은 청소년기 약물 사용과 알로스태틱 부하 수준을 순차적으로 높여 심혈관 질환 위험을 유의하게 높이는 것으로 나타났다. 본 연구는 아동기 역경 경험이 성인기 신체 건강에 미치는 부정적 영향의 생물학적 기제를 밝히고, 심혈관 질환 예방에 있어서 가정환경적 요인과 개인의 행동적 요인의 통합적인 접근의 중요성을 강조한다. 더불어, 본 연구는 아동기 역경 경험의 예방 및 맞춤형 대책 수립을 위한 과학적 근거를 제공하였음에 의의를 가진다. This study examined the longitudinal impact of adverse childhood experiences (ACEs) on the risk of cardiovascular disease (CVD) in adulthood, mediated through adolescent substance use (SU) and allostatic load (AL). We used the Add Health dataset, a nationally representative sample in the United States. The data spanned from Wave 1 (childhood, ages 12~17) to Wave 4 (adulthood, ages 25~32). The study included 3,254 participants who had complete data for major variables (ACEs, SU, AL, and CVD), were not pregnant in Wave 4, and were adolescents in Wave 2 (ages 13 ~ 17) (1,380 males and 1,874 females). Adverse childhood experience questions were assessed in Waves 1 and 3, while adolescent substance use questions were measured in Wave 2. Allostatic load and risk of cardiovascular disease questions were obtained in Wave 4. Control variables included biological sex, age, race, and childhood adversity experience at Wave 3. Descriptive statistics and correlation analysis were conducted using SPSS 25, and a structural equation model was analyzed through Mplus 8.7. The primary findings of this study are as follows: First, it was found that adverse childhood experiences, adolescent substance use, and allostatic load directly and positively influenced the risk of cardiovascular disease in adulthood. Second, adverse childhood experiences were found to positively influence the risk of cardiovascular disease by sequentially increasing adolescent substance use and allostatic load. The findings of this study provide a specific mechanism for how adverse childhood experiences impact physical health problems in adulthood. Moreover, it underscores the need for integrating environmental and behavioral factors when addressing cardiovascular disease prevention. This study is significant as it identifies one of the physiological mechanisms that affect the risk of cardiovascular disease in adulthood.

인슐린 농도와 심혈관질환과의 환자-대조군 연구

김연경 연세대학교 보건대학원 2006 국내석사

이 연구는 혈중인슐린 농도와 심혈관질환 발생기전의 이해를 도모하고, 더불어 삶의 질 향상 및 의료비 절감, 국민건강증진에 기초 자료로 사용하고자 수행하였다.$$a$$a연구대상은 2001년 2월~2004년 12월 연세대학교 의료원 심혈관질환 유전체 연구에 참여한 환자군 2,586명, 대조군 1,514명으로 하였으며, 연구결과는 다음과 같다.$$a$$a심근경색에 영향을 미치는 위험요인 중 성별(남성)이 가장 영향이 컸으며, 성별과 연령이 통제된 상태에서 고밀도지단백, 공복 시 혈당, 저밀도지단백, 허리둘레, 중성지방, 인슐린저항성 지수, 혈중인슐린 농도가 주요관련요인으로 밝혀졌다.$$a$$a고혈압에 영향을 미치는 위험요인 중 허리둘레가 7.2(%)으로 가장 영향이 컸으며 다음으로는 성별(남성, 2.2%), 중성지방/고밀도지단백, 공복 시 혈당, 저밀도지단백, 혈중인슐린 농도 순이었다.$$a$$a관상동맥질환에 영향을 미치는 위험요인 중 허리둘레가 20.8(%)으로 가장 영향이 컸고 고밀도지단백이 5.5(%)으로 두번째로 영향이 컸으며 저밀도지단백, 공복 시 혈당, 연령, 성별, 체질량지수, 중성지방, 혈중인슐린 농도, 중성지방/고밀도지단백 순이었다.$$a$$a혈중인슐린 농도를 다섯 단계로 나눈 후 나이, 성별, 인슐린, 중성지방, 중성지방/고밀도지단백, 저밀도지단백, 공복 시 혈당, 체질량지수를 보정한 후 혈중 혈중인슐린 농도에 따라 심혈관질환 발생 위험을 살펴보았다. 혈중인슐린 농도(0-5µU/ml)에 비해 혈중인슐린 농도(7.1-9µU/ml)에서 심혈관질환에 걸릴 위험이 1.331배 높았다. 또한 혈중인슐린 농도 (9.1-10µU/ml)에서도 혈중인슐린 농도(0-5µU/ml)에 비해 심혈관질환에 걸릴 위험이 1.830배, 마지막으로 혈중인슐린 농도(12.1-97.4µU/ml)에서 혈중인슐린 농도(0-5µU/ml)에 비해 심혈관질환에 걸릴 위험이 1.856배 높았다.$$a$$a결론적으로, 혈중인슐린 농도가 높아질수록 심혈관질환에 걸릴 위험성이 점차 증가하였으며 통계학적으로 상당히 유의하였다. 고인슐린혈증은 기타 인자들과는 독립적으로 심혈관질환 발생을 예측할 수 있었다.$$a$$a The aim of this study is to provide correlations between blood insulin level and cardiovascular disease. It will help understanding of the mechanism of cardiovascular disease, improve quality of life, reduce medical expenses, and be used as basic data for health promotion of a nation.$$a$$aThe objects of study are 2,586 cases and 1,514 controls who participated in the Yonsei University Cardiovascular Genome Center(Seoul, Korea) from Feb 2001 to Dec 2004. The followings are the results.$$a$$aIt was verified that myocardial infarction was most affected by gender(male). It was also ascertained that high density lipoprotein, fast blood sugar, low density lipoprotein, waist, triglyceride, homeostasis model assessment of insulin resistance, and insulin level is the main factor for myocardial infarction.$$a$$aThe most risk factor which affects hypertension was waist(7.2%). The next most risk factor was sex(male 2.2%). Triglyceride/high, density lipoprotein, fast blood sugar, low density lipoprotein, and insulin level was the next orders.$$a$$aThe most risk factor which influences coronary artery disease was waist(20.0%). the next most risk factor was high density lipoprotein(5.5%), low density lipoprotein, fast blood sugar, age, sex, body mass index, triglyceride, insulin level, triglyceride/high, density lipoprotein were the next orders.$$a$$aInsulin level was divided by five level and age, sex, triglyceride, triglyceride/high, density lipoprotein, low density lipoprotein, fast blood sugar, and body mass index are controled to examine risk of cardiovascular disease according to insulin level. it was identified that cardiovascular disease would be taken 1.331 times higher when the level of insulin was 7.1-9µU/ml than 0-5µU/ml and when the level of insulin was 9.1-10µU/ml cardiovascular disease could be taken 1.830 times higher compared with the level of insulin 0-5µU/ml. At last when the level of insulin was 12.1-97.4µU/ml cardiovascular disease could be gotten 1.856 times higher than 0-5µU/ml.$$a$$aIn conclusion, as level of insulin increases higher, a possibility of danger to be affected by cardiovascular disease grows. In other words there is the intimate relation between insulin level and cardiovascular disease. It was estimated that the hyperinsulinemia can cause cardiovascular disease without other effects independently.$$a$$a

A Study on the Relationship of Riboflavin and Niacin with Cardiovascular disease

Ahn, Ji Yeong Graduate School, Eulji University 2019 국내석사

Cardiovascular disease can be the result of different factors: Innateness factors such as aging, family history, and sex; biological factors such as high blood pressure and cholesterol level; or environmental factors such as smoking, eating, and stress. The Korean National Health and Nutrition Examination consists of health and nutrition surveys and the examination of related research. This study examined the factors associated with cardiovascular disease and ways to reduce its prevalence by analyzing the data within the Korean National Health and Nutrition Examination reports from 2013 to 2017, in particular, survey and examination data for people over 50 years old. Multiple associations between myocardial infarction, angina, hypertension, and hyperglycemia diagnoses and lipid level- and waist size-related variables were analyzed to examine the correlations between cardiovascular disease and physiological factors. Factor-related risk levels were then determined through logistic regression analysis. To study the correlations between physiological factors and a standard rate of nutrition intake, adherences to the recommended dietary allowances of riboflavin and niacin were examined. The study population was divided into two groups: members of the first group consumed riboflavin and niacin at the recommended rate while those in the second group did not. The results showed hypertension, hyperglycemia, abnormal waist size, and a low high-density lipoprotein (HDL) -cholesterol level were correlated with the presence of cardiovascular disease (p<0.001). Compared with the normal group, hypertensive individuals had 1.2 times, those with hyperglycemia had 1.6 times, those with an abnormal waist size had 1.7 times, and those with a low HDL cholesterol level had 1.6 times higher prevalence of cardiovascular disease. Riboflavin intake was correlated with hypertension, hyperglycemia, and low HDL-cholesterol level, whereas niacin intake was correlated with hypertension, hyperglycemia, waist size, and low HDL-cholesterol level. The combination of niacin and riboflavin intakes was correlated with hypertension, hyperglycemia, waist size, and a low HDL-cholesterol level (p<0.05). The combined intake of the recommended levels of riboflavin and niacin reduced the average physiological factor abnormality rate to 80%. The significant relationship between cardiovascular disease and the recommended intakes of riboflavin and niacin (p<0.001) shows the importance of individual volition. Among individuals who took riboflavin at the standard rate, cardiovascular disease prevalence was reduced to 64%, in those who did so for niacin the prevalence rate was 63%, and in those who did so for both riboflavin and niacin, the prevalence rate was markedly reduced to 50%. In conclusion, consumption of riboflavin and niacin at their recommended rates can affect an individual’s physiological components, such as hypertension, hyperglycemia, waist size, and HDL-cholesterol level. The beneficial effects of riboflavin and niacin can reduce cardiovascular disease prevalence and, importantly, can reduce the death rate associated with cardiovascular disease.

(A) study on health-log application for self-diagnosis and prophylaxis of cardiovascular disease

Jo, Sangil Sungkyunkwan university 2019 국내석사

In this paper, we propose a novel application in the form of life log for the prevention and improvement of cardiovascular disease. The proposed application can recognize user's face and estimate user’s current cardiovascular state in real time. This study also recommends personalized content for prevention and improvement of cardiovascular disease based on measured results and user's body information. Users can check their status anytime from anywhere. They can also use recommended content to effectively prevent and improve cardiovascular disease. Major parameters of cardiovascular disease such as pulse rate, blood pressure, pulse wave velocity, and arrhythmia are estimated using signal processing techniques and deep learning techniques. Personalized content recommendations provide cardiovascular prevention and remediation through music and food, such as food intake guide based on user's pulse rate and target pulse rate through sound source that could induce normal pulse rate, and calorie estimation through food recognition based on diagnosis results and user's body information.

노인요양시설 요양보호사의 뇌심혈관계 질환 예방관리를 위한 참여형 개선기법(PAOT) 적용 효과

정윤주 연세대학교 보건대학원 2019 국내석사

목적 : 본 연구에서는 뇌심혈관질환의 위험인자를 가지고 있는 노인요양시설 요양보호사를 대상으로 참여형 개선기법(Participatory Action Oriented Training, PAOT)을 적용한 뇌심혈관질환 관리 효과를 분석해 보고자 하였다. 대상 및 방법 : A지역 근로자건강센터의 ‘요양보호사를 위한 건강관리 프로그램’에 참여한 노인요양시설 요양보호사 중 뇌심혈관질환 발병 위험인자를 가지고 있고 뇌심혈관질환 발병 위험도를 평가결과 ‘최저위험군’이상인 51명에 대하여 참여형 개선기법(PAOT)의 원리를 적용하여 뇌심혈관계질환 예방 관리 프로그램을 실시하였으며, 같은 기간 동안 동일조건과 동일 프로그램에 참여한 요양 보호사 중 참여형 개선기법을 적용하지 않은 82명을 대조군으로 선정하여 뇌심혈관질환 발병 위험인자 및 발병 위험도 평가 결과를 프로그램 전∙후 및 대조군과의 차이를 비교 분석하였다. 참여형 개선기법(PAOT)은 임상검사와 생활습관 설문을 통하여 뇌심혈관질환 위험인자 및 뇌심혈관질환 발병 위험도 평가를 평가하며, 체크리스트 작성을 통한 본인의 문제 인지, 담당간호사와의 토의식 면담, 개선 계획의 작성, 개선 개획 실천에 대한 격려, 개선 계획의 실천 확인, 프로그램 결과 평가로 구성되어있다. 임상검사 결과를 바탕으로 본인의 유험요인에 대해 확인하도록 하였고, 생활습관 설문(체크리스트)을 통해 ‘잘하고 있는 점’과 ‘개선이필요한 점’을 인지시켰다. 결과 : 프로그램에 참여한 51명이 작성한 개선 계획은 총 146개로 1명당 2.9개의 개선계획을 세웠고, 3개월 후 개선계획의 실천여부를 확인한 결과 총 85개의 개선계획을 실천하고 있어 58.2%의 실천율을 보였다. 프로그램 후 대상자의 흡연율은 13.9% 감소하였고, 규칙적인 운동습관은 37.3% 증가하였으며, 과음주군은 17.6%유의하게 감소하였다. 프로그램 후 대상자의 허리둘레는 1.2cm감소하였고, 수축기 혈압(SBP) 9.3mmHg, 이완기 혈압(DBP) 4.6mmHg이 유의하게 감소하였다. 공복시 혈당은 8.67㎎/㎗감소하였고, 중성지방 수치도 유의하게 감소하였다. 뇌심혈관질환 발병 위험도 평가 결과 중제군에서 개선된 경우는 23명(45.1%)이었으며, 대조군은 18명(22.0%)으로 통계학적으로 유의한 차이가 있었다(p<0.001). 결론: 노인요양시설 요양보호사 51명을 대상으로 참여형 개선기법(PAOT)을 적용한 뇌∙심혈관계 질환 예방관리 프로그램을 진행한 결과 흡연, 음주, 신체활동 지수가 개선되었고, 이를 통해 뇌심혈관질환 위험요인인 혈압, 혈당, 중성지방, 허리둘레가 개선되어 45.1%의 대상자의 뇌심혈관질환 발병 위험도가 감소되는 효과가 있었다. Effects of Participatory Action Oriented Training(PAOT) on Prevention and Management of Cerebral Cardiovascular Disease for Elderly Caregivers in Elderly Nursing Care Facilities. The purpose of this study was to evaluate the effectiveness of the program by administering the principles of Participatory Action Oriented Training(PAOT) for the elderly caregivers in elderly nursing care facilities with risk factors for cerebral cardiovascular disease. The subjects of this study were 51 elderly caregivers in elderly nursing care facilities with risk factors for cerebral cardiovascular disease. The program for prevention and management of cerebral cardiovascular disease using the participatory improvement technique(PAOT) evaluates the risk of cerebral cardiovascular disease risk and cerebrovascular disease risk through clinical examination and lifestyle questionnaire. Knowing your health problems through the creation of checklists, discussion with the nurse in charge, creating a lifestyle plan, support for action, planning practice verification, and program evaluation. After three months, we checked whether changes in lifestyle such as smoking, drinking, physical activity were changed and blood pressure, blood sugar, blood cholesterol, obesity and waist circumference were measured. The risk level was re-evaluated and compared with the control group. A total of 146 improvement plans were made by 51 people who participated in the program, 2.9 improvement plans were made per one person, and 58.2% of them were implemented. three months later. After the program, the smoking rate of the subjects decreased by 13.9%, regular exercise habits increased by 37.3%, and the frequency of alcohol consumption decreased by 17.6%. In the clinical study, the waist circumference of the subject was decreased by 1.2cm, the systolic blood pressure(SBP) was 9.3mmHg, and the diastolic blood pressure(DBP) was decreased by 4.6mmHg, Fasting blood sugar decreased by 8.67mg/dl and triglyceride level was also significantly decreased. Evaluation results of the risk of developing cerebral cardiovascular disease were 23(45.1%) in the intervention group and 18(22.0%) in the control group, which were statistically significant differences(P<0.001). As a result of advancing a prevention and management program for cerebral cardiovascular diseases applying participatory action oriented training(PAOT) to 51 elderly caregivers in elderly nursing care facilities, smoking, excessive drinking state and exercise life practice rate improved. And blood pressure, blood glucose, neutral fat and abdominal circumference, which are risk factors for cerebral cardiovascular disease, were improved, and 45.1% of the subjects had the effect of reducing the risk of developing cerebral cardiovascular disease.

Cardiovascular risk assessment and machine learning prediction model of metabolic comorbidity

안서경 서울대학교 대학원 2022 국내박사

연구 배경: 인구의 고령화와 서구형 생활양식으로 인해 대사 질환 동시 이환 (고혈압, 당뇨병, 및 고지혈증 등을 포함한 두가지 이상의 대사 질환을 가진 것으로 정의)의 유병률이 증가하고 있다. 이러한 대사성 질환은 심혈관계 질환의 위험 증가와 연관된다. 2016년 Global Burden of Disease에 따르면, 심혈관계 질환에 의한 사망은 21세기 주요 사망 원인이며, 우리나라에서는 암에 이어 두번째로 높은 사망원인을 차지한다. 세계보건기구 (The World Health Organization)에서는 음주, 흡연, 비만, 신체 활동, 건강한 식습관을 심혈관계 질환의 예방 가능한 요인으로 지정한 바 있다. 이에 대사 질환 동시 이환에 대한 연구가 필요하다. 따라서, 이 연구의 목적은 1) 한국에서의 대사성 질환과 동시 이환의 유병률을 추정하고; 2) 대사 동시 이환 심혈관계 가족력과 심혈관계 발생 위험을 평가하고, 3)대사 동시 이환에 따른 심혈관계 사망에 대해 생활습관 요인 미치는 영향을 평가하고; 4) 생활 습관 변화와 대사 증후군의 연관성을 확인하고; 5) 대사 동시 이환에 대한 기계학습을 기반으로 한 건강 연령 및 질병 위험 예측 모형을 개발하는 것이다. 연구 방법: 본 연구는 한국인유전체역학조사사업 (KoGES)의 도시기반 (Health examinee-Gem Study, HEXA), 농촌기반 (Cardiovascular disease association study, CAVAS), 지역사회기반 (Ansan and Ansung Study, 2001-2014)를 주로 사용하였고, 추가로 미국 국민건강영양조사 (US National Health and Nutrition Examination Survey, NHANES 2003-2014), 한국국민건강영양조사 (Korea NHANES, KNHANES 2007-2014), 아시아 코호트 연구 (Asia Cohort Consortium)를 사용하였다. 통계방법으로는, 세계보건기구의 세계표준인구를 이용한 직접 표준화 방법을 이용해 대사성 질환의 연령표준화 유병률을 산출하였다. 연구 대상자의 일반적인 특성은 연속형 변수의 경우 Student’s t-test, 범주형 변수의 경우 Chi-squared test를 시행하여 비교하였다. 콕스 비례 위험 회귀 분석과 로지스틱 회귀 분석을 수행하여 hazard ratios (HRs), odds ratio (ORs), 95% confidence interval을 추정하였다. 위험 예측 모형의 경우, training set (전체 대상자의 70%)에서 콕스 비례 회귀 분석, random survival forest 기반 모형을 각각 구축하고, test set (전체 대상자의 30%)에서 concordance index (c-index)를 이용해 각 모형의 성능을 평가하였다. 건강 연령 예측 모형의 경우, 10-fold validation을 사용한 elastic net 방법을 이용해 모형을 구축하였다. 연구 결과: 한국과 미국의 대사성 질환과 동시 이환을 비교한 결과, 한국이 미국보다 대사 동시 이환의 유병률이 낮았다. 한국과 미국에서 가장 흔한 대사 질환 조합은 고혈압과 비만이었다. 한국 인구 중 농촌 지역에 거주하는 인구는 도시 지역에 거주하는 인구보다 대사 동시 이환 유병률이 더 높은 것으로 나타났다. 대사 동시 이환, 심혈관계 질환 가족력, 그리고 심근경색과 뇌졸중을 포함한 심혈관계 질환의 위험 연구 결과는 다음과 같다. 고혈압, 당뇨병, 고지혈증이 있고, 심혈관계 가족력이 있는 대상자는 심혈관계 질환 가족력과 질병이 없는 대상자에 비해 유의하게 심혈관계 질환 (HR 2.88, 95% CI: 1.96-4.24), 심근경색 (HR 3.30, 95% CI: 2.06-5.29), 뇌졸중 (HR 2.52, 95% CI: 1.33-4.79) 위험이 증가하는 것을 확인했다 심혈관대사 질환 동시 이환을 가진 대상자에서 생활 습관 요인이 심혈관계 질환 관련 사망에 미치는 영향 연구에서는, ‘비흡연’, ‘금주’, ‘체질량 지수 18.5–27.4kg/m2’를 건강 상태로 정의하여 건강한 생활 습관 점수를 산출했다. 생활 습관 요인 중 금연은 심혈관계 질환 사망 위험 감소와 가장 강한 연관성을 보였다. 고혈압, 당뇨병, 관상동맥질환이 있는 대상자에서는 건강한 생활 습관 점수가 1씩 증가할 때마다 심혈관계 사망위험이 24% (HR 0.76, 95% CI: 0.63-0.93)씩 감소했다. 2개 이상의 심혈관계 대사질환이 있는 대상의 경우, 건강한 생활 습관 요인은 3가지 모두 있는 경우 심혈관계 질환 사망 (HR 0.51, 95% CI: 0.42-0.61)과 심혈관계 질환으로 인한 조기 사망위험(HR 0.38, 95% CI: 0.27-0.54)의 감소에 유의한 영향이 있었다. 지역사회기반 연구자료를 이용한 반복 측정된 생활 습관 요인의 변화에 따른 대사 증후군 위험 연구에서는, 하루 흡연 개피수의 증가 (HR 1.49, 95% CI: 1.09-2.03), 음주량의 light/moderate에서 heavy로 증가는 (HR 1.42, 95% CI: 1.10-1.84) 대사 증후군의 발생 위험의 증가와 유의한 연관성을 보였다. 새롭게 비만 된 대상자는 꾸준히 적정 체중을 유지하는 대상자에 비해 대사성 증후군 (HR 1.88, 95% CI: 1.44-2.45)의 발생 위험의 증가와 유의한 관계를 보였다. 보다 정밀한 개인 맞춤 건강 상태 예측 및 개선을 위해 기계 학습 기반 질병 예측 모형을 개발과 대사 동시 이환에 대한 예측 변수로서의 건강연령을 개발한 연구에 따르면, 실제 연령에 비해 젊은 건강 연령을 가진 경우, 당뇨병 (HR = 0.63, 95% CI: 0.55–0.72), 고혈압 (HR = 0.74, 95% CI: 0.68–0.81), 당뇨병과 고혈압 동시 이환 (HR = 0.65, 95% CI: 0.47–0.91) 위험도가 낮은 것으로 나타났다. 기계학습기반 예측 모형 연구 결과, 기계 학습 기반의 고혈압과 당뇨병 동시 이환 모형은 높은 통계적 질병 예측력을 보이는 것으로 나타났다. 연구 결론: 본 연구는 한국 인구집단에서 심혈관계 질환 발생 및 사망의 위험을 줄이기 위해 대사 동시 이환에 대한 연구에 대한 필요성을 강조한다. 본 연구에서는 동시 이환을 가진 대상자 중 특히 심혈관계 질환 가족력이 있는 경우에 심혈관계 질환의 발생 위험이 증가하는 것을 확인하였다. 또한 심혈관계 대사 질환 동시 이환을 가진 대상자라도, 금연, 금주, 표준 체질량 지수 유지와 같은 건강한 생활 습관은 심혈관계 질환으로 인한 사망과 조기 사망 위험 감소와 연관성이 있었다. 또한, 건강한 생활습관으로의 변화를 통해 대사 증후군의 위험을 줄이는 데 도움이 되는 것을 확인하였다. 이러한 요인들을 기반으로 기계학습을 이용하여 구축된 질병 예측 모형과 건강연령은 우리나라에서의 대사 질환 동시 이환에 대한 고위험군을 파악하고 이를 미리 예방함으로써, 건강증진을 통해 질병 부담을 줄이는 효과적인 도구로 활용될 수 있을 것으로 기대된다. Introduction: The growing aging population and westernized lifestyle have increased the prevalence of disease comorbidity, which is defined as having more than two metabolic diseases including hypertension (HTN), diabetes mellitus (DM), dyslipidemia (LIP), obesity, and metabolic syndrome (MetS). The combination of these diseases is related to an increased risk of cardiovascular disease (CVD) outcomes. The Global Burden of Disease 2016 Study reported that CVD are by far the leading cause of death globally and one of the major health challenges of the 21st century. In Korea, CVD is the second largest cause of death following cancer. As those diseases share risk factors, the World Health Organization (WHO) designated healthy lifestyle, including alcohol reduction, weight loss, smoking cessation, physical activity, and healthy diet, as modifiable factors of CVDs. Thus, it is necessary to estimate the amount of comorbidity prevalence, identify the combined association of metabolic comorbidity and other risk factors (family history of CVD and lifestyle factors) with CVD outcomes, and develop predictive model for comorbidity for detecting the high-risk of metabolic comorbidity and preventing the future risk of CVD through intervention strategies. Methods: This study mainly used population-based cohort study from the Korea Genome and Epidemiology Study (KoGES) including Health Examinee-Gem study (HEXA), cardiovascular disease association study (CAVAS), and Ansan and Ansung Study from 2001-2014, in addition to United States (US) National Health and Nutrition Examination Survey 2003-2014 (NHANES), Korea NHANES (KNHAENS) 2007-2014, and Asia Cohort Consortium (ACC) study. For the statistical analyses, direct standardization methods using the WHO world standard population was performed to estimate the age-standardized prevalence of metabolic diseases. The baseline characteristics were compared using Chi-squared test for categorical variables and Student’s t-test for continuous variables. Cox proportional hazards regression analysis was performed to estimate hazard ratios (HRs) with 95% confidence intervals (CIs) of CVD outcomes. To calculate the odds ratios (ORs) of metabolic diseases, logistic regression models were used. For prediction model, cox proportional hazard regression, and random survival forest (RSF) models were developed in the training set (70% of the total population) and performance evaluations of each model were performed in the test set (30% of the total population) with concordance statistics (c-index). For self-assessed biological age (BA) prediction model, elastic net regression analysis with 10-fold cross validation was performed. Results: According to the comparison of the prevalence of metabolic disease and comorbidity in Korea and the US, Korea had a lower prevalence of metabolic comorbidity than the US. In both Korean and the US population, the most common combination was HTN and obesity. Among the Korean population, individuals living in rural areas had the higher comorbidity prevalence than those who lived in urban areas. In the association between metabolic comorbidity, family history of CVD, and the risk of CVD study, we found that individuals with DM, HTN, LIP, and a positive family history of CVD had a 2.88-fold increased risk of CVD, a 3.30-fold increased risk of MI, and a 2.52-fold increased risk of stroke compared to the individuals with a negative family history of CVD and none of metabolic diseases. In the impact of lifestyle factors with cardiometabolic disease (CMDs) such as HTN, DM, coronary heart disease (CHD), and stroke on CVD death study, the healthy lifestyle status was defined as ‘never smoker’, ‘never drinker’, and ‘body mass index (BMI) 18.5–27.4kg/m2’in Asian population. Among the lifestyle factors, non-smoking had the strongest association with decreasing risk of all cause and CVD death among the healthy lifestyle factors. A significant association of healthy lifestyle score with lower CVD death was observed among individuals with HTN, DM, and CHD (HR 0.76, 95% CI: 0.63-0.93). For individuals with cardiometabolic comorbidity, having three of healthy lifestyle factors was significantly associated with decrease in CVD (HR 0.51, 95% CI: 0.42-0.61) and premature CVD death (HR 0.38, 95% CI: 0.27-0.54). Based on the repeated measurements for assessing change in lifestyle factors study, unhealthy lifestyle modification including increased dose of cigarette smoking (HR 1.49, 95% CI: 1.09-2.03) and increased their intensity of consumption from light/moderate to heavy had a significantly increased risk for MetS (HR 1.42, 95% CI: 1.10-1.84). For obesity, individuals who newly became obesity had a significant increase in risk for MetS (HR 1.88, 95% CI: 1.44-2.45). For improving the individualized health status, we developed machine learning-based disease prediction model and self-assessed BA as a predictor for metabolic comorbidity. We found that compared to the individuals in same BA as chronological age (CA) group, those in younger BA than CA group were associated with a decreased risk of DM (HR = 0.63, 95% CI: 0.55–0.72), HTN (HR = 0.74, 95% CI: 0.68–0.81), and combination of HTN and DM (HR = 0.65, 95% CI: 0.47–0.91). For machine learning-based disease prediction model study, predictive models achieved a high discriminatory ability for comorbidity of HTN and DM. Conclusions: This study highlights the necessity of accounting to metabolic comorbidity to reduce the future risk of CVD outcomes in Korean population. Although individuals already have had cardiometabolic comorbidity, healthy lifestyles (smoking cessation, abstaining from alcohol, and maintaining BMI) are effective to reduce the further risk of CVD death. Moreover, lifestyle changes help to decrease the risk of a cluster of metabolic conditions. At last, machine learning-based self-assessed BA and disease prediction model may be an effective indicator for identifying the high-risk group and decreasing burden of metabolic comorbidities in Korea through prevention.

Mechanical and Biomaterials Engineering Optimization of Complex Cardiovascular Operations and Innovative Therapeutics

Zhu, Yuanjia Stanford University ProQuest Dissertations & These 2023 해외박사(DDOD)

Cardiovascular disease is the leading cause of death globally. Every year, close to 18 million people succumb to complications of cardiovascular disease. Coronary artery disease and valvular diseases, two of the main types of cardiovascular disease, are both highly prevalent and frequently treated. For valvular heart disease in particular, guidelines continue to evolve to advocate for earlier intervention. Overall, this represents a significant economic burden with over $200 billion each year being spent on cardiovascular disease in the United States.Severely diseased valves require surgical intervention to prevent life-threatening complications. Even though valve replacement has been a mainstay of therapy for decades, choosing between a mechanical or bioprosthetic replacement valve involves committing patients either to lifelong anticoagulation or future reoperation for structural valve deterioration. Valve repair on the other hand is a safe and highly effective strategy that avoids the disadvantages associated with valve replacement. The repaired valve has its native tissue preserved, and it can heal, grow, adapt, and resist infection. New guidelines also continue to advocate for valve repair over replacement whenever possible. However, valve repair techniques have been based primarily upon anatomic principles and guided by appearance and function. The biomechanical foundations for these techniques are unknown and represent a significant unmet need. In fact, a lot of the repair techniques were developed by trial and error. As such, early failure after valve repair still occurs, and long-term repair durability can still be improved. The cardiothoracic surgery department's clinical practice has notoriety in valve repair with high volume of clinical substrate. This provides us with a wealth of opportunity to study this topic.In terms of ischemic cardiovascular disease, the primary therapy focus has been on urgent revascularization via percutaneous coronary intervention or coronary artery bypass grafting. Although these techniques restore macrovascular blood flow, they do not address microvascular malperfusion, metabolism, and dysfunction, all of which can result in cardiomyocyte loss, maladaptive left ventricular remodeling, progression to heart failure, and early mortality. This is a highly significant clinical problem, and there is a huge unmet need to bring the missing substrate back to the ischemic tissues.Our lab developed an innovative 3D-printed left heart simulator which allowed us to use biomechanical engineering tools to investigate valvular disease and analyze surgical repair treatment while collecting quantitative data. Building on this heart simulator technology foundation, this dissertation first details novel aortic and mitral valve disease models and repair techniques, followed by device development for aortic and mitral valve diseases. Finally, by harnessing the power of 3D bioprinting and photosynthetic biologic agents, I summarized photosynthetic biologic therapeutics for the treatment of ischemic cardiovascular disease. The research outlined herein has resulted in a significant clinical impact on intracardiac valve repair and novel therapeutics for ischemic cardiovascular diseases. This work will continue to serve as a foundation for future investigations of clinical therapies with the goal of rapid and safe translation to patient care.