随机森林与支持向量机在预测乌鲁木齐农村老年人养老服务需求的应用研究

目的 探讨随机森林(random forest, RF)和支持向量机(support vector machine, SVM)算法在农村老年人养老需求服务预测中的应用价值。方法 运用分层整群抽样对乌鲁木齐1652名农村老年人进行问卷调查,采用RF算法和SVM算法构建模型预测农村老年人的日常生活照料、医疗保健服务、精神慰藉服务和休闲娱乐服务等四类养老服务需求,通过准确率、特异度、召回率和ROC曲线进行模型评估。结果 日常生活照料、精神慰藉服务需求中RF模型ROC曲线下面积更大;医疗保健服务需求中SVM模型的ROC曲线下的面积更大;休闲娱乐服务需求中两个模型的ROC曲线下面积相等。结论 RF算法和SVM算法构建的老年人养老服务需求预测模型,各有其优势,在养老事业的精准预测工作中有一定的价值。     

基于KNN和RF算法构建危重患者全因死亡预测模型的研究

目的评估机器学习算法对单中心危重患者全因死亡率的预测性能。方法采用美国大型重症医学数据库MIMIC-III作为数据来源,建立基于机器学习的K最近邻(KNN)和随机森林(RF)开发评估重症患者预后的预测模型,并与传统的简化急性生理评分-II(SAPS-II)进行比较。使用Boruta进行特征变量的筛选,通过交叉验证对模型性能进行评估,主要研究结局为危重患者90d死亡率。结果研究纳入30833例患者,SAPS-II评分[中位数(四分位间距)]为33(25~43),其中5534例(17.9%)患者在进入ICU后90d内死亡。3种模型得到的AUC值分别为:SAPS-II模型0.75±0.03,KNN模型0.75±0.03,RF模型0.80±0.03。结论与传统的SAPS-II评分相比,机器学习算法中的RF模型可以更精准地预测危重患者的预后。     

应用深度神经网络和Klemera-Doubal方法估计生物学年龄

目的 将深度神经网络(DNN)和Klemera-Doubal方法(KDM)应用于中国中老年人群估计生物学年龄(BA)，并选择最优方案评价模型表现。方法 从中国健康与养老追踪调查(CHARLS)2011—2012年的全国基线调查数据中，选取45岁≤年龄≤85岁中老年人的血液标志物样本(11 513人)、身体测量指标样本(13 603人)和血液+身体测量样本(9 904人)，应用DNN和KDM估计BA，根据Pearson相关系数、平均绝对误差(MAE)和均方根误差(RMSE)评价BA的估计准确性和选出最优方案。通过计算BA和时序年龄(CA)回归的残差ΔBA，与死亡情况、认知能力构建复杂抽样logistic回归或线性回归模型。结果 3种样本均是血液+身体测量样本计算的BA与CA相关性最高(r DNN=0.91;r KDM=0.48),MAE(MAE_DNN=3.11;MAE_KDM=13.74)和RMSE(RMSE_DNN=4.15;RMSE_KDM=17.76)最低，DNN的BA估计准确性优于KDM。在血液+身...     

利用机器学习算法分析影响阿尔兹海默病的脂蛋白及代谢物

目的 基于机器学习算法探讨阿尔兹海默病发病的脂蛋白及代谢物影响因素。方法 从ADNI数据库中选取2012年诊断结果为正常（cognitive normal,CN）和阿尔兹海默病（Alzheimer disease,AD）的研究对象共314例,收集其脂蛋白及代谢物数据。采用随机森林、lasso回归、XGboost算法三种方法对变量进行重要性排序及筛选。利用三种方法筛选出的变量,结合研究人群的性别、年龄、婚姻状况构建随机森林模型,预测影响AD发病的重要因素。结果 三种方法共筛选出12个脂蛋白及代谢物变量,结合研究人群的年龄、性别、婚姻状况共15个变量被纳入随机森林模型。模型的准确率为84.13%、灵敏度为93.75%、特异度为53.33%、Kappa值为0.518 3、AUC(95%CI)为0.735(0.600～0.871)。根据随机森林模型中Mean Decrease Accuracy和Mean Decrease Gini两指标分别筛选出的排名前五的变量中均包含以下四个变量：大极低密度脂蛋白中的磷脂与总脂质之比（L＿VLDL＿PL＿PCT）、年龄（AGE）、乳糜微粒和极大极低密度脂蛋白...     

基于奇异谱分析的ARIMA模型在山西省流感预测中的应用

目的 分析基于奇异谱分析（singular spectrum analysis, SSA）的自回归移动平均模型（Autoregressive integrated moving average, ARIMA）模型预测流感样病例 (influenza like illness, ILI) 发病趋势的可行性，为流感防控工作提供合理的预测方法。 方法 利用山西省2010年第14周-2017年第13周的流感监测资料以不同长度配比的训练集、测试集构建SSA-ARIMA模型，并与ARIMA、BP神经网络（Back propagation neural network, BPNN）、广义回归神经网络（General Regression Neural Network, GRNN）模型进行比较。采用平均绝对误差（Mean Absolute Error，MAE）、均方误差（Mean Squared Error，MSE）、均方根误差（Root Mean Squared Error，RMSE）比较各模型预测效果。 结果 模型拟合方面，SSA-ARIMA模型在预测未来一个月发病趋势时的MAE、MSE、RMSE分别为0.163、0.061、0.248；预测六个月时分别为0.161、0.061、0.248；预测一年时分别为0.168、0.066、0.256；均低于ARIMA、BPNN、GRNN。模型预测方面，在预测未来一个月发病趋势时的MAE、MSE、RMSE分别为0.056、0.005、0.068；预测六个月时分别为0.189、0.081、0.285；预测一年时分别为0.210、0.075、0.273；也均低于ARIMA、BPNN、GRNN。 结论 SSA-ARIMA模型对山西省ILI的预测效果优于ARIMA、BPNN、GRNN，可为流感预测提供科学依据。     

集成学习在水产品无机砷含量预测中的应用研究

目的水产品中无机砷元素慢性暴露严重危害人体健康,其检测时间成本较高,为实现无机砷元素含量快速预测,构建一种基于小样本特征量的无机砷元素含量预测集成学习模型。方法抽取2018—2022年秦皇岛地区水产品重金属元素检测数据,采用皮尔逊法对铅、镉、汞、无机砷4种元素做相关性分析并检验多重共线性。使用逐步回归向量组合法测试不同特征组合下梯度提升回归树(gradient boosting regressor,GBR)和随机森林模型(random forest,RF)的拟合优度(R²)及均方误差(mean squared error,MSE),以筛选最优组合。综合比对5种集成学习算法在模型评估指标、预测误差比率分布、目标危险系数(target hazard quotients,THQ)三方面的预测效果以评估方法可行性。结果4种重金属元素间呈弱相关,不存在多重共线性。RF及GBR算法拟合优度R²分别为89.9%、93.3%。极端学习树模型(extremely randomized tree,ET)在贝、鱼、蟹、虾类水产品中R²分别为1...     

生物学年龄评价方法进展

利用生物学年龄衡量个体的衰老进程,可以弥补实足年龄难以解释同龄个体间健康水平差异的缺陷。近年来,涌现出多种基于临床或表型、分子生物学指标或复合型指标的生物学年龄测量方法。本文对一些常见的生物学年龄评价指标进行概述,并比较这些评价指标的效度。     

阻塞性睡眠呼吸暂停综合征对成人哮喘程度的影响

目的 评估阻塞性睡眠呼吸暂停综合征(obstructive sleep apnea syndrome, OSAS)对成人支气管哮喘(bronchial asthma, BA)程度的影响，分析二者之间的潜在关联性。方法 采用回顾性研究收集2018年1月～2022年6月入江苏省中医院进行治疗的90例BA患者为研究对象，所有患者均接受多导睡眠监测(polysomnography, PSG),以此计算呼吸暂停低通气指数(apnea-hypopnea index, AHI)。依据AHI值将患者分为轻度OSAS(AHI<15)、中度OSAS(1530)。此外，测量患者BA相关指标如第一秒用力呼气量(FEV1)、FEV1%、FEV1/用力肺活量(FVC)、呼出气一氧化氮(FeNO)、总气道阻力(R5)、中心气道阻力(R20)以及周边弹性阻力(X5)。采用方差分析比较不同程度OSAS患者BA相关指标差异，Pearson线性相关分析AHI与BA指标测量值的相关性。结果 AHI<15组为1例(1.11%),15相似文献   

基于SMOTE算法与机器学习的老年人健康素养预测研究

目的 探讨合成少数类过采样技术(SMOTE)结合机器学习模型在老年人是否具备健康素养预测评估中的应用。方法 利用单因素筛选从资料中筛选出与是否具备健康素养有关联的变量;以筛选出的变量作为输入变量,以是否具备健康素养为结局变量,分别在经SMOTE算法处理前后的数据集中建立logistic回归模型、随机森林和SVM模型,通过受试者工作特征曲线(ROC)来评价模型性能。结果 Logistic回归、随机森林和SVM在SMOTE算法处理前的测试集中的准确率分别为0.833、0.600和0.636,3种模型的ROC曲线下面积(AUC)分别为0.723、0.815和0.728;在SMOTE算法处理后的测试集中的准确率分别为0.936、0.908和0.890,3种模型的AUC分别为0.896、0.944和0.897。结论 随机森林模型在老年人是否具备健康素养的预后评估中具有较高的应用价值。     

多项目联合检测在类风湿性关节炎诊断及活动度鉴别中的价值研究

目的 探讨淋巴细胞亚群、免疫球蛋白游离κ链、游离λ链、补体C3、C4以及其他常用检测指标在类风湿性关节炎(RA)诊断及其活动度判断中的价值。方法 选取2020年3月—12月在金华市中心医院住院的215例RA患者作为患者组,根据DAS28评分将RA患者分为活动期组(127例)和稳定期组(88例),同时期149例健康体检者作为正常组,比较研究组和正常组、RA活动期组和稳定期组类风湿因子(RF)、抗环瓜氨酸多肽抗体(抗CCP)、红细胞沉降率(ESR)、C-反应蛋白(CRP)、淋巴细胞亚群以及免疫球蛋白游离κ、λ链、补体C3、补体C4等指标水平,并采用ROC曲线评估其对RA的诊断和判别RA活动期的价值。结果 研究组215例,正常组149例,研究组Th细胞、Ts细胞、游离κ链、游离λ链、抗CCP、ESR、CRP、RF高于正常组,差异有统计学意义(P <0.05)。活动期组的年龄、Td细胞、C3、游离κ链、游离λ链、ESR、CRP等指标均高于稳定期组,差异有统计学意义(P <0.05)。当RF的临界值为16.5 U/ml时,AUC为0.993,敏感度为80.0%,特异度为94.6%,诊...     

Construction of an integral formula of biological age for a healthy Chinese population using principle component analysis

Background

Whereas chronological age (CA) cannot distinguish functional differences among individuals of the same age, the biological age (BA) may be used to reflect the functional state of the body. The purpose of this study was to construct an integral formula of the BA, by using principle component analysis (PCA).

Methods

The vital organ function of 505 healthy individuals of Han origin (age 35–91 years) was examined. A total of 114 indicators of cardiovascular, pulmonary, and brain functions, and clinical, inflammatory, genetic, psychological, and life habit factors were assessed as candidate indicators of aging. Candidate indicators were submitted with CA to correlation and redundancy analyses. The PCA method was used to build an integral formula of the BA for the population.

Results

Seven biomarkers were selected in accordance with a certain load standard. These biomarkers included the trail making test (TMT), pulse pressure (PP), mitral valve annulus ventricular septum of the peak velocity of early filling (MVES), minimum carotid artery intimalmedial thickness (IMTmin), maximum internal diameter of the carotid artery (Dmax), maximal midexpiratory flow rate 75/25 (MMEF75/25), and Cystatin C (CysC). The formula for the BA was: BA = 0.0685 (TMT) + 0.267 (PP)–1.375 (MVES) + 22.443 (IMTmin) + 2.962 (Dmax)–2.332 (MMEF75/25) + 16.104 (CysC) + 0.137 (CA) + 0.492.

Conclusion

Several genetic and lifestyle indicators were considered as candidate markers of aging. However, ultimately, only markers reflecting the function of the vital organs were included in the BA formula. This study represents a useful attempt to employ multiple indicators to build a comprehensive BA evaluation formula of aging populations.     

Comparison of visually estimated age with physiologically predicted age as indicators of rates of aging

The commonly held view that people age at different rates derives largely from visual estimates of age. Although most people ‘look their age’ everyone can cite examples of individuals in middle and late adulthood who appear to be aging very slowly or very rapidly. Efforts to quantify aging rates scientifically require measurement of a large number of physiological parameters in a large population sample. This paper compares visual estimates of age with physiologically predicted measures to determine their value as indicators of the rate of aging.This study used data from 1086 male participants in the Baltimore Longitudinal Study of the Gerontology Research Center, NIA. These men have provided comprehensive biomedical and psychosocial data at one and one-half year intervals for as long as 20 years. The visual estimate of age was made by the examining physician at the first study visit of each participant, without knowledge of the man's actual age. The error of this estimate was determined by subtracting actual age from estimated age. Correlation analysis of error in estimated age with an objective assessment of biological age based on physiological variables indicated a significant association between the two approaches. When men who have died since their study participation were compared with survivors, the former were found to have been significantly ‘older for their age’ than the latter using both visual and physiological estimate approaches. To determine whether certain lifestyle traits were associated with variation in these two indicators, multiple regression analyses were performed. These showed that men who smoked, who are fatter, or who were in poor health were predicted as older than their chronological age peers using both approaches. Results of this study suggest that the easily determined visual estimate of age may be a useful indicator of aging rate within a population.     

Biological value of the average daily diets of elderly and old individuals in Uzbekistan

Nutrition was studied in the elderly (60-74 years), old individuals (75-90 years), and long livers (above 90 years) of Tashkent and the Kashkadarya Region by estimating the biological values of their diets and the biochemical parameters of vitamin C metabolism. Low biological values of actual diets, low vitamin C supply were found in the elderly and old individuals. Alimentary correction of the biochemical parameters of vitamin C metabolism was ascertained to be possible by enhancing the biological value of their diets by taking into account the national peculiarities of nutrition and relevant dietary technologies.     

Gender disparity of successful aging in Taiwan

The purpose of this longitudinal study was to examine the association of gender on successful aging during a 6-year follow-up study. Data were drawn from the 1989, 1993, 1996, and 1999 Survey of Health and Living Status of the Elderly in Taiwan, a national representative population survey of 4,049 participants aged 60 years or older; 43% were women. Successful aging indicators were defined as physical (absence of disease and physical disability), mental (normal cognitive function and absence of depression), and social (good social support and participation in productive activities). Gender differences related to successful aging during the 6 years were analyzed with Cox proportional hazards regression model. Significant disparities were found between elderly men and women in activities of daily living (ADL), instrumental activities of daily living (IADL), cognitive function, depressive symptoms, social support and productive activities. When the data included individuals who were deceased, elderly women showed better maintenance of ADL and less vulnerability to IADL impairment than elderly men. However, when the analysis included only survivors during the follow-up, the elderly were more likely to be impaired in IADL and cognitive function, have depressive symptoms, and lack engagement in productive activities. Cumulative disadvantage was associated with health disparities in elderly women. This research demonstrates the need to develop gender-sensitive policies.     

Gender inequalities in health among older Brazilian adults.

OBJECTIVE: To investigate gender differences among older Brazilians in their health status and their use of health services. METHODS: Participants were individuals aged 60 years and older included in a national household survey conducted in Brazil in 1998. Data were analyzed by multiple logistic regression, taking into account the design effect due to multistage sampling. RESULTS: There were differences in the health and living conditions of older men and older women that were not explained by age or place of residence. Older women had worse indicators of schooling and personal income but better indicators of housing standards and per capita household income. The older women also reported more chronic diseases, had poorer indicators of independence and physical mobility, sought health services more often, and reported more medical visits in the previous year. Despite their apparent worse health conditions, elderly women in urban areas had lower hospitalization rates in the previous year (odds ratio = 0.89; 95% confidence interval, 0.82-0.96) than did elderly men in urban areas. CONCLUSIONS: Our results indicate that among older Brazilians there are gender inequalities in health that cannot be explained by age and place of residence. The findings raise questions on how health, socioeconomic, and cultural factors influence gender patterns of seeking and using health care in later life in the country. As pressures on health care and health funding increase in Brazil as a result of the aging of the population, there is a need to take a gender perspective into account.     

On the empirical association between poor health and low socioeconomic status at old age

Epidemiologic studies using mortality rates as indicators of health fail to find any meaningful association between poor health and low socioeconomic status in older age-groups, whereas economic studies using self-assessed health consistently find a significant positive correlation, even after controlling for self-reporting errors. Such contradictory results have not been reported for working age individuals. A simple explanation might be that the elderly samples on which the epidemiologic and economic studies are based come from different populations. However, this paper shows that similar contradictory results are obtained even when the same samples are used, simply by switching between self-assessed health and mortality as health indicators. An alternative explanation is proposed, namely that these health indicators yield different results because they relate to different ranges of the latent health variable at old age.     

Importance of Considering Malnutrition and Sarcopenia in Order to Improve the QOL of Elderly Hemodialysis Patients in Japan in the Era of 100-Year Life

In the current aging society of Japan, malnutrition and resultant sarcopenia have been widely identified as important symptomatic indicators of ill health and can cause impairments of longevity and quality of life in older individuals. Elderly individuals are recommended to have sufficient calorie and protein intake so as to enjoy a satisfactory quality of life, including maintaining activities of daily living in order to avoid emaciation and sarcopenia. The prevalence of emaciation and sarcopenia in elderly hemodialysis (HD) patients in Japan is higher than in non-HD elderly subjects due to the presence of malnutrition and sarcopenia associated with chronic kidney disease (CKD). Furthermore, comorbidities, such as diabetes and osteoporosis, induce malnutrition and sarcopenia in HD patients. This review presents findings regarding the mechanisms of the development of these early symptomatic conditions and their significance for impaired QOL and increased mortality in elderly HD patients.     

西部地区宁夏城乡老年人健康自我评价的对比研究

摘要:目的　了解西部宁夏地区老年人健康自评状况,分析影响老年人健康自评因素的城乡差异。方法　宁夏地区60岁及以上的城市和农村老年人各1 501名作为研究对象,内容包括社会人口学资料、健康自评、社会交往意愿、养老意愿、社会服务满意度等。结果　宁夏地区城乡老年人健康自评结果比较差异无统计学意义。年龄、外出频次是城乡老年人的共同影响因素,高龄、外出频次≤2次的老年人健康自评较差。另外,城市老年人健康自评还与社区服务满意度相关（χ2＝3.160,P＜0.001）,对社区服务满意的老年人要比不满意的健康自评好;农村老年人健康自评还与养老意愿相关（χ2＝9.658,P＜0.001）;有居家养老意愿的老年人比机构养老意愿的健康自评好。结论　西部宁夏地区老年人健康自评整体情况没有明显城乡差异,健康自评影响因素有一定不同。护理人员应根据实际情况为老年人提供不同的卫生服务,才能有效地提高老年人群的健康水平。     

Nutritional intake in both free-living and institutionalized older adults with xerostomia

A common problem in the institutionalized as well as the free-living elderly is inadequate nutrition. This problem will continue in significance as the population continues to age. Previous studies have indicated a variety of nutritional deficiencies in the diets of older adults. Chronic degenerative diseases and usage of multiple pharmacologic agents are also prevalent in the geriatric population and confound this complex problem. A common observation among the many contributing factors to the problem of relative or absolute malnutrition in seniors is that many individuals exhibit the sign/symptom of xerostomia (dry mouth). Recent studies indicate that nearly one in five older adults exhibit xerostomia. Xerostomia increases the difficulty of these older adults in obtaining the proper nutrition. Problems in lubricating, masticating, tolerating, tasting and swallowing food contribute significantly to the complex physiological and psychological manifestations of aging. To this date the literature fails to demonstrate an association between xerostomia and malnutrition in the elderly.