GRU和GBDT混合模型在早产风险预测中的应用

早产是新生儿死亡及病残的首要原因, 且影响新生儿的远期健康. 然而早产的准确预测一直是医学上的一个难题. 目前医学上早产的早期筛查多基于特殊检查, 但因成本核算等问题难以大规模临床应用, 而电子病历的普及和人工智能技术的发展, 为产科疾病的早期风险评估提供支持. 本文利用产科电子病历的诊疗信息, 构建GRU和GBDT的...     

Use of Computer Interview Data to Test Associations between Risk Factors and Pregnancy Outcomes

Computer interviews have been used in a variety of settings as a means of gathering data and providing health education information. The objective of this study was to determine whether data gathered from a computer interview have predictive validity in determining pregnancy outcomes. Pregnant women (N= 190) completed a computer-assisted interview to provide risk factor information. Medical records were reviewed to obtain prenatal and birth outcome information. Twenty-nine percent experienced prenatal complications and nearly half experienced problems related to labor and delivery. After known risk factors were controlled for, self-reported psychosocial problems, low social support, and substance abuse by the father were associated with pregnancy complications. An unbalanced diet and low social support were associated with lower infant birth weight. Findings suggest that data collected via computer interviews can provide useful risk factor screening information.     

Applying data mining to explore the risk factors of parenting stress

High parenting stress has been shown to be associated with illness, poor marital relationships, and child abuse. It is thus important to detect and reduce the stress early. Numerous situational, socioeconomic, child, and parent factors contribute to parenting stress. However, regression analysis, the traditional method of exploring risk factors in medicine and social science, has the limitation of not showing the classification, nor exploring unknown potentional factors. Hence, the goal of this study is to explore the risk factors of parenting stress using data mining with decision tree C5.0, to obtain more information. The data are from a professional research group, TBPS, in the National Taiwan University. A total of 206 mother-term born child dyads were recruited to complete the measures of the Parenting Stress Index (PSI), the Child Behaviour Checklist (CBCL), the Comprehensive Developmental Inventory for Infants and Toddlers (CDIIT), and the Chinese Toddler Temperament Scale (CTTS), and so this database includes thousands of variables. The study results indicate that a child development problem, CDIIT, is the major contributing factor to parents with the highest stress, the 90% parenting stress group. For the 80%, 70%, and 60% parenting stress groups, the behavioural problem of children, CBCL, is the major factor causing parenting stress. The data mining decision tree showing the classification route of risk factors is better than the regression model at detecting the significant factors. The findings in this work are considered helpful references for medical staff and social workers to help parents prevent and reduce their parenting stress and thus promote health.     

基于产检数据聚类的妊娠合并症可视分析

妊娠合并症通常是指在尚未怀孕或妊娠期间由其他原因导致的疾病。很多情况下,孕妇在产检时各项检查结果表现正常,但却在之后的妊娠期间罹患妊娠合并症,所以单凭单项指标并不能很好的发现潜在的妊娠合并症。通过选取多组孕妇的多维产检指标数据进行聚类分析,在此基础上设计并开发了探索妊娠合并症潜在可能性的对比可视分析系统。案例分析表明,临床妇科医生能够通过可视化界面观察就诊孕妇妊娠期间的生理指标,查找与之相似病例,与传统就诊方式相比,提高发现其患有潜在妊娠合并症的概率已达到74.8%,平均误差率降低2.8%。     

Use of Chaos Theory and Complex Systems Modeling to Study Alcohol Effects on Fetal Condition

A systems dynamics computer model to predict birth complications for individual pregnant woman was developed from prospectively conducted data on a database of 125 pregnant women. The model is based upon nonlinear mathematics derived from the study of chaos and complex systems. The model was then tested prospectively on 27 additional pregnant women, making predictions on their level of obstetrical risk. The model was refined until it correctly predicted the outcomes of all 125 cases in the development database. Prediction was made with an accuracy of 25/27 cases for the prospective test cases. Predictions were made for fetal condition at birth, presence or absence of operative delivery, and presence or absence of uterine dysfunction. Then the model was used to explore alcohol use during pregnancy. A reasonable spread or alcohol use existed among subjects, allowing consideration of alcohol effects. Alcohol was found to have differential effects on fetal condition at birth depending upon the presence or absence of high levels of psychosocial stress and the use or other substances. In all cases, the effect of alcohol was only evident after the 10 drinks per week level was reached. For the high-stress/one other substance group, there could be an 18-fold effect on fetal condition at birth. For the low-stress/one other substance group, the effect was only 3-fold, and for the alcohol alone group, the effect was negligible.     

A Computer Program for Estimating Individualized Probabilities of Breast Cancer

A computer program is presented for estimating the absolute risk of developing breast cancer over a specified age interval for women with combinations of five risk factors, namely age at menarche, age at first live birth, family history of breast cancer, number of previous breast biopsies, and presence of atypical hyperplasia in biopsy specimens. Statistical methods have been developed and applied to data from the Breast Cancer Detection and Demonstration Project to obtain (i) point estimates of absolute risk by combining relative risk estimates from case-control data and estimated composite incidence rates from cohort data, and (ii) confidence intervals by using implicit delta-method arguments. The program is interactive and easy to use and is therefore well suited to assist in medical counseling. For instance, women with high estimated risk might be advised to undergo a program of frequent surveillance with mammography. An executable version for IBM-compatible PCs is available from the author upon submission of a PC diskette formatted with MS-DOS.     

Back pain of working pregnant women: Identification of associated occupational factors

The objective of this study was to identify major occupational factors that were significantly correlated with back pain in pregnant women working in higher education, health care and service areas. A total of 73 working pregnant women were surveyed using questionnaires specifically designed for evaluating correlations between occupational factors and severity of back pain; 37 women were interviewed at both 20 and 34 weeks of pregnancy, 17 at 20 weeks only, and 19 were interviewed at 34 weeks only. “Rest breaks allowed” and “job autonomy” were negatively correlated with severity of back pain at 20 weeks of pregnancy. “Staying in a confined area” and “having restricted space” were positively correlated with severity of back pain at 34 weeks of pregnancy. The study suggests that allowing pregnant women to take more rest breaks and to have more job autonomy may reduce the severity of back pain during early pregnancy, and that allowing movement outside the working area and providing less restricted space may reduce back pain during late pregnancy.     

Performance of physical tasks in pregnancy.

Two hundred women in the third trimester of a first pregnancy were interviewed, and 50 of them were interviewed again at 4 months post partum. They were asked whether they experienced greater current difficulty in performing 46 common tasks than they did before pregnancy and were asked to grade their responses. Additional information concerning specific factors affecting each activity, such as discomfort and fatigue, was obtained, together with data concerning socioeconomic status and stress perception. The data were analysed to identify the activities whose performance was most and least affected by pregnancy, indicating that the pregnant condition affects perceived difficulty of performance in many activities. Although some activities are not amenable to ergonomic intervention, the study highlighted some in which equipment design and selection could improve the match between a task and the pregnant.     

Postpartum depression prediction through pregnancy data analysis for emotion-aware smart systems

Emotion-aware computing represents an evolution in machine learning enabling systems and devices process to interpret emotional data to recognize human behavior changes. As emotion-aware smart systems evolve, there is an enormous potential for increasing the use of specialized devices that can anticipate life-threatening conditions facilitating an early response model for health complications. At the same time, applications developed for diagnostic and therapy services can support conditions recognition (as depression, for instance). Hence, this paper proposes an improved algorithm for emotion-aware smart systems, capable for predicting the risk of postpartum depression in women suffering from hypertensive disorders during pregnancy through biomedical and sociodemographic data analysis. Results show that ensemble classifiers represent a leading solution concerning predicting psychological disorders related to pregnancy. Merging novel technologies based on IoT, cloud computing, and big data analytics represent a considerable advance in monitoring complex diseases for emotion-aware computing, such as postpartum depression.     

Research on lower body shape of late pregnant women in Shanghai area of China

Anthropometric data, especially the data of lower body of late-pregnant women, has become an immediate need for ergonomic design of maternity products, such as clothing, related motherhood supplies, maternity support products, and so on. So 55 late pregnant women in Shanghai area of China were randomly selected for this lower body shape research. By anthropometric measurement method 18 dimensions were measured to characterize lower body shape of late pregnant women. Mean values, standard deviations (SD), coefficients of variation, and percentiles for each variable were estimated. It was found that the morphology of hip and abdomen mostly influenced the lower body shape. Then according to the characters of hip and abdomen, the late pregnant women's lower body shapes were divided into two types. Moreover this paper also analyzed the change trend of lower body shape along late pregnancy months (7 to 9th months). As a result of this research, the late pregnant women's lower body shapes were described from different aspects which were thought to provide a guide for the maternity products design.Relevance to industryThe anthropometric data is essential to design reasonable and professional maternity products for late pregnant women, such as maternity support products, maternity clothing and so on. Especially this study can benefit the maternity garment industry. Moreover utilization of the anthropometric data, product designers would be able to outfit to a wider range of target users.     

A comparison of sitting posture adaptations of pregnant and non-pregnant females

The postural differences between seated pregnant and non-pregnant subjects were studied, and the effects of various sitting workstation designs were assessed. Five pregnant and five non-pregnant subjects participated in this study. The experimental variables were the table-chair-vertical-distance (35 cm and 30 cm) and the seat-inclination-angle (0°, +10° and +20°). The seated working postures of the last 10 min of a 50-min sitting session were videotaped and analyzed using a 2D motion analysis system. Postural differences were observed between the pregnant and non-pregnant groups which may be attributed to the changes and/or mechanisms associated with pregnancy. The increased size of the abdomen requires more room in front of the body; in the pregnant subjects this was obtained by holding the upper-arm further from the body, inclining the upper trunk more anterior, and extending the hip more. A seat with a sloping angle of 10° was found to reduce the biomechanical loading on the lower back and in subjective ratings from both groups, was preferred over the other seat angles.

Relevance to industry

Pregnant women make up a small but significant part of the workforce. Changes in body form and working posture due to pregnancy change the effective working envelope and increase the biomechanical loads on the musculoskeletal system resulting in an increase in the risk of lower-back pain. This study examined the posture adaptations and the effects of workstation design on pregnant and non-pregnant subjects.     

Evidence-based personal applications of medical computing models in risk factors of cardiovascular disease for the middle-aged and elderly

Medical applications on cardiovascular disease (CVD) for hybrid computing models are an emerging research area. The CVD, including stroke, hypertension, and high cholesterol, is one of 10 leading causes of death in Taiwan in middle-aged and elderly; in particular, the CVD has become the top killer in advanced countries. Thus, this serious but interesting issue triggers the study to focus on patients of the CVD. The study explores variables, influencing cardiovascular functions for four risk factors of blood pressure, blood glucose, blood fat, and kidney diseases, in the middle-aged and elderly. By the data collection of regular physical examination system from a regional hospital, the original dataset contains 52 variables collected from October 2011 to February 2014. We model a hybrid knowledge-based classification system to organize expert experiences, integrated linear and nonlinear attribute selection methods, data discretization of smart expert method, rough set theory, the LEM2 algorithm, and rule-filtering technique to classify the CVD for the early warning purpose. After data cleaning, 20 attributes with 2027 records are remained. For effectively identifying the variables of CVD subjects, this study reclassifies the above four risk diseases into three classes: no disease, 1&2 diseases, and 3&4 diseases. To verify performance of the proposed procedure, we experience an empirical experiment to compare the full 20 used attributes, the used attributes of integrated linear and nonlinear attribute selections with rule-filtering technique, and various classifiers. Conclusively, the 13 used attributes obtained from optimal accuracy become the key determinants that affect the four risk factors of the CVD. The empirical results and findings benefit doctors’ and medical institutions’ early medical recommendations and treatments with the advantages of significantly reducing morbidity of CVD.     

基于LSTM与XGBOOST混合模型的孕妇产后出血预测

孕妇产后大出血是造成全球孕妇死亡的重要因素之一,在我国位居孕妇死亡原因首位,然而对产后出血的提前判定一直以来都是医学上一个难题.电子病历的普及,以及机器学习和深度学习技术的发展,为预知孕妇产后大出血提供了基于大数据的解决办法.本文提出利用孕妇的电子病历数据,构建基于LSTM和XGBoost的混合模型来预测孕妇产后大出血.实验结果表明,利用基于LSTM和XGBoost的混合模型对孕妇产后大出血进行预测是可行的,能够为医生判断孕妇产后出血情况提供参考,为孕妇分娩时是否需要备血方案提供决策支持,对降低产后大出血致死率具有积极意义.     

Evolutionary computation in the identification of risk factors. Case of TRALI

This paper presents the use of an evolutionary algorithm hybridized with the concepts of testor and typical testor in determining factors associated with transfusion related acute lung injury (TRALI). Although nowadays many cases of this syndrome remain ignored or misdiagnosed, this is the leading cause of morbidity and mortality related to transfusion in the United States.This research was conducted with data from 174 cases collected in the Centenary Hospital Miguel Hidalgo in the city of Aguascalientes, Mexico, in the period 2007 to 2010.The proposed algorithm works with information from the model known as “two hits”, in which the first hit is the original disease and the second corresponds to the blood transfusion. This algorithm was strengthened with mechanisms that let it do an efficient search in the whole solution space. In addition to the calculation of the informational weight, the algorithm also establishes the cutoff point that determines the variables that impact the most.From the results given by the algorithm and the cutoff proposed by the medical staff, a strategy for the treatment of patients that should be transfused was proposed.This study confirmed some of the risk factors previously reported in the literature, and also made an interesting discovery.     

Trunk postural demands of occupational activities of some merchant pregnant women in Benin,West Africa

Strenuous physical work puts expectant mothers at risk of experiencing back pain during the gestational months. Pregnant women in Benin perform physically demanding occupational tasks that include the lifting and carriage of heavy loads on their heads for commercial activities. A large percentage of pregnant subjects (58%) reported having back pain episodes since the start of their pregnancy. However, the mean Oswestry score of the affected participants was relatively low with a mean score of 0.2 (SD: 0.12), on a scale form 0 to 1. An evaluation of the postural demands of the occupational activities of these women revealed that they performed on average 328 trunk flexions at angles exceeding 60°, with 66 of these flexions sustained for more than 4 s, during the average 7.9 h where trunk postures were recorded. They also spent on average 36% of the recording time at trunk flexion angles larger than 20°. These results show that the merchant pregnant women in the Porto-Novo area in Benin are at great risk for developing back disorders during pregnancy.

Practitioner Summary: Results will make a first contribution to the literature by identifying the stressful postures adopted during a typical day. The findings of this study can help in the development of preventative concepts and postural modification techniques to decrease the occurrence of back pain during pregnancy for women in Benin.     

Biomechanics of rising from a chair and walking in pregnant women

The present study aimed to assess the changes in the pattern of rising from a chair and walking forward as pregnancy progressed. Twelve pregnant women and 10 nulliparous women were included in this study. Participants were videotaped with a digital video camera in the sagittal plane, and the coordinates of the markers attached to the subjects were identified using image analysis software. The peak trunk-flexion angle in pregnant women during rising was smaller, but the hip-extension angle during the stance phase was larger than in controls. Also, the peak horizontal and vertical velocities of the center of mass were lower, and appeared earlier, in pregnant women than in controls. During rising, pregnant women dampened the propulsion attributable to increased uterus volume, and they enhanced the forward propulsion at gait initiation. To ensure safe motion, pregnant women should not initiate gait until reaching a stable standing position after rising.     

Bayesian non-linear regression models with skew-elliptical errors: Applications to the classification of longitudinal profiles

Typically, the fundamental assumption in non-linear regression models is the normality of the errors. Even though this model offers great flexibility for modeling these effects, it suffers from the same lack of robustness against departures from distributional assumptions as other statistical models based on the Gaussian distribution. It is of practical interest, therefore, to study non-linear models which are less sensitive to departures from normality, as well as related assumptions. Thus the current methods proposed for linear regression models need to be extended to non-linear regression models. This paper discusses non-linear regression models for longitudinal data with errors that follow a skew-elliptical distribution. Additionally, we discuss Bayesian statistical methods for the classification of observations into two or more groups based on skew-models for non-linear longitudinal profiles. Parameter estimation for a discriminant model that classifies individuals into distinct predefined groups or populations uses appropriate posterior simulation schemes. The methods are illustrated with data from a study involving 173 pregnant women. The main objective in this study is to predict normal versus abnormal pregnancy outcomes from beta human chorionic gonadotropin data available at early stages of pregnancy.     

Bayesian non-linear regression models with skew-elliptical errors: Applications to the classification of longitudinal profiles

Typically, the fundamental assumption in non-linear regression models is the normality of the errors. Even though this model offers great flexibility for modeling these effects, it suffers from the same lack of robustness against departures from distributional assumptions as other statistical models based on the Gaussian distribution. It is of practical interest, therefore, to study non-linear models which are less sensitive to departures from normality, as well as related assumptions. Thus the current methods proposed for linear regression models need to be extended to non-linear regression models. This paper discusses non-linear regression models for longitudinal data with errors that follow a skew-elliptical distribution. Additionally, we discuss Bayesian statistical methods for the classification of observations into two or more groups based on skew-models for non-linear longitudinal profiles. Parameter estimation for a discriminant model that classifies individuals into distinct predefined groups or populations uses appropriate posterior simulation schemes. The methods are illustrated with data from a study involving 173 pregnant women. The main objective in this study is to predict normal versus abnormal pregnancy outcomes from beta human chorionic gonadotropin data available at early stages of pregnancy.