核偏最小二乘回归及其在医学中的应用

目的研究用于处理解释变量与反应变量之间非线性关系或复杂关系的一种基于核函数的回归方法:核偏最小二乘回归。方法运用Monte-Carlo模拟方法,对核偏最小二乘回归的模型拟合效果和预测效果予以分析。结果模拟试验结果表明:核偏最小二乘回归估计性能均较高。结论核偏最小二乘回归是基于核函数的非线性回归方法,模型构建基于样本,而非解释变量空间,该方法特别适合于处理医学研究中各种类型资料,能够有效地处理解释变量与反应变量之间的非线性关系或复杂关系等方面。     

偏最小二乘回归模型在高校科技人员类型预测中的应用

目的探讨某省高校科技人员类型的影响因素,建立偏最小二乘回归模型,对高校科技人员类型进行预测、判别,为今后高校科技人员的培养及其合理使用提供更加可靠、科学的理论依据。方法采用自编调查量表——高校科技人员影响因素调查表对该省内高校科技人员进行随机抽样调查,采用Chronbachsα系数和因子分析对量表的信度和效度进行检验,运用SAS9.1对收集的数据进行偏最小二乘回归分析。结果 Chronbachsα系数=0.781,表明调查表具有较好的内在一致性信度,因子分析结果显示量表同时具有较高的结构效度。偏最小二乘回归结果显示,该模型具有较好的拟合优度,并符合专业上的解释,可为人事、科研和教育部门进一步完善高校科技人员培养机制提供理论参考依据。结论偏最小二乘回归模型作为一种新兴的统计分析方法适合应用于高校科技人员类型预测的研究。     

偏最小二乘判别分析在基因微阵列分型中的应用

目的由于疾病,特别是肿瘤的识别模型,其分型准确度对疾病的治疗和预后具有重要意义,因而,本研究探讨了基于基因表达谱的疾病分型识别模型建模方法。方法结合白血病基因表达谱数据分析,利用偏最小二乘判别分析(PLS-DA)对利用基因微阵列数据予以建立白血病分型模型,并与Golub等提出的建模方法相对照,比较它们的判别效果。结果基于偏最小二乘判别分析的白血病识别模型的拟合准确度和预测准确度均达到100%。结论研究表明,基于偏最小二乘判别分析的模型明显提高了白血病的分型正确率,无论是拟合精度,还是预测精度,均高于Golub等提出的方法。     

厦门市女性乳腺癌死亡与减寿趋势分析

目的探讨厦门市女性乳腺癌死亡和减寿的变化趋势,为厦门市乳腺癌综合防治工作提供依据。方法收集整理2009-2014年厦门市女性乳腺癌死亡资料计算死亡率、平均减寿年数(AYLL)、死亡率年均变化百分比等评价指标,用GM(1,1)模型对死亡率和AYLL进行预测。结果 2009-2014年,厦门市女性乳腺癌死亡率7.10/10万,死亡年龄中位数为56岁,死亡率随年龄的升高而升高;造成的AYLL为16.59年。GM(1,1)模型对AYLL和PYLLR的预测值与实际值平均相对误差为5.92%和2.45%,预测2015-2017年女性乳腺癌所致AYLL和PYLLR值有所上升。结论 GM(1,1)模型可用于厦门市女性乳腺癌减寿趋势预测,未来厦门市女性乳腺癌死亡年龄有年轻化趋势,应重视乳腺癌的预防控制工作。     

用最小—乘法预测婴儿死亡率

对我市10年的婴儿死亡率,用最小一乘法建立了对数曲线模型,y=22.8-9.3806Lgx,用于婴儿死亡率的预测。结果优于最小二乘法。理论值与实际值的相关系数r为0.8079,并预测我市1995年婴儿死亡率达12.3‰,比1990年下降16.3％;2000年婴儿死亡率为10.4‰,比1990年下降29.3％。     

核正交偏最小二乘在代谢组学数据分析中的应用

目的探索核正交偏最小二乘方法的特点及其在代谢组学数据分析中的应用。方法通过模拟实验和真实代谢组学数据,评价核正交偏最小二乘方法的模型预测能力及其可视化效果。结果模拟数据分析表明,当数据间存在线性关系时,KOPLS与传统的线性OPLS具有相同的效果;当数据间存在非线性关系时,KOPLS具有相对更高的预测能力,得分图的可视化效果更好。实际数据分析结果显示,应用KOPLS能够提高模型预测能力和改善可视化效果。结论对于高维非线性关系的代谢组学数据更适合使用KOPLS方法。     

应用灰色系统GM（1,1）模型预测恶性肿瘤死亡率

灰色系统GM(1,1)模型是根据过去和现实的信息建模,推断将来的情况,提出事物发展变化的规律。本文利用该模型将鞍钢职工1983～1987年恶性肿瘤死亡率加以整理,建模,并预测了1988和1990年的恶性肿瘤死亡率为207.45和234.70/10万。文中又用最小二乘法的预测数据来验证灰色系统GM(1,1)模型预测的精度。通过分析和预测结果看到鞍钢职工恶性肿瘤死亡率呈逐年上升趋势。     

用量小—乘法预测婴儿死亡率

对我市１０年的婴儿死亡率，用最小一乘法建立了对数曲线模型，ｙ＝２２．８－９．３８０６Ｌｇｘ，用于婴死亡率的预测。结果优于最小二乘法。理论值与实际值的相关系数ｒ为０．８０７９，并预测我市１９９５年婴儿死亡率达１２．３‰，比１９９０年下降１６．３％；２０００年婴儿死亡率为１０．４‰，比１９９０年下降２９．３％。     

中国城市化对儿童死亡率的影响

目的:探讨中国2000～2008年各地区城市化对儿童死亡率的影响。方法:基于W.Henry Mosley&Lincoln C.Chen提出的儿童死亡率决定因素的理论分析框架,构建城市化对儿童死亡率影响的路径模型,选取2000～2008年各地区的有关统计资料,借助偏最小二乘路径模型技术(PLS path modeling)进行实证分析。结果:城市化促进了经济发展,经济发展提高了基础设施水平,改善了通信条件,加大了污染控制力度,提高了人民收入;上述因素促进了妇幼保健服务利用,降低了儿童营养不良的发生率,最终降低儿童死亡率。结论:在中国各地大力推进城市化的背景下,控制污染、提高人民收入、改善通信条件、发展文化教育事业、加强基础建设,有助于降低儿童死亡率。     

成都市人口死亡率时间序列分析与预测

目的采用时间序列分析和预测成都市人口死亡率的动态发展趋势,建立时间序列模型,考察模型的应用效果并做出预测。方法利用时间序列自相关系数和偏相关系数识别模型,采用最小二乘法估计模型参数,用Box-Ljung统计量评价ARIMA模型的拟和度,用平均预测相对误差作为预测效果的评价指标。结果建立乘积ARIAM(0,1,1)(0,1,1)12模型,模型平均绝对百分误差MAPE=8.50%。成都市人口死亡率自2000年逐渐下降,预计序列后2年将继续呈现下降趋势。结论所运用的时间序列分析和预测模型拟合效果较好,可应用于疾病发病和死亡动态变化规律的分析和其未来发展趋势的预测、预报。     

Addressing the identification problem in age-period-cohort analysis: a tutorial on the use of partial least squares and principal components analysis

In the analysis of trends in health outcomes, an ongoing issue is how to separate and estimate the effects of age, period, and cohort. As these 3 variables are perfectly collinear by definition, regression coefficients in a general linear model are not unique. In this tutorial, we review why identification is a problem, and how this problem may be tackled using partial least squares and principal components regression analyses. Both methods produce regression coefficients that fulfill the same collinearity constraint as the variables age, period, and cohort. We show that, because the constraint imposed by partial least squares and principal components regression is inherent in the mathematical relation among the 3 variables, this leads to more interpretable results. We use one dataset from a Taiwanese health-screening program to illustrate how to use partial least squares regression to analyze the trends in body heights with 3 continuous variables for age, period, and cohort. We then use another dataset of hepatocellular carcinoma mortality rates for Taiwanese men to illustrate how to use partial least squares regression to analyze tables with aggregated data. We use the second dataset to show the relation between the intrinsic estimator, a recently proposed method for the age-period-cohort analysis, and partial least squares regression. We also show that the inclusion of all indicator variables provides a more consistent approach. R code for our analyses is provided in the eAppendix.     

Selenium in forage crops and cancer mortality in U.S. counties

The potential protective effect of selenium status on the risk of developing cancer has been examined in animal and epidemiologic studies. This ecological study investigated the association between U.S. county forage selenium status and site- and sex-specific county cancer mortality rates (1950-1969) using weighted least squares regression. Consistent, significant (p less than .01) inverse associations were observed for cancers of the lung, rectum, bladder, esophagus, and cervix in a model limited to rural counties and for cancers of the lung, breast, rectum, bladder, esophagus, and corpus uteri in a model of all counties. No consistent significant positive associations were observed in the rural county models. This remarkable degree of consistency for the inverse associations strengthens the likelihood of a causal relationship between low selenium status and an increased risk of cancer mortality.     

Double robust estimation of optimal partially adaptive treatment strategies: An application to breast cancer treatment using hormonal therapy

Precision medicine aims to tailor treatment decisions according to patients' characteristics. G-estimation and dynamic weighted ordinary least squares are double robust methods to identify optimal adaptive treatment strategies. It is underappreciated that they require modeling all existing treatment-confounder interactions to be consistent. Identifying optimal partially adaptive treatment strategies that tailor treatments according to only a few covariates, ignoring some interactions, may be preferable in practice. Building on G-estimation and dWOLS, we propose estimators of such partially adaptive strategies and demonstrate their double robustness. We investigate these estimators in a simulation study. Using data maintained by the Centre des Maladies du Sein, we estimate a partially adaptive treatment strategy for tailoring hormonal therapy use in breast cancer patients. R software implementing our estimators is provided.     

Modeling the impact of comorbidity on breast cancer patient outcomes

The objective of this paper is to model the impact of comorbidity on breast cancer patient outcomes (e.g., length of stay and disposition). Previous studies suggest that comorbidities may significantly affect mortality risks for breast cancer patients. The 2006 AHRQ Nationwide Inpatient Sample (NIS) is used to analyze the relationships among comorbidities (e.g., hypertension, diabetes, obesity, and mental disorder), total charges, length of stay, and patient disposition as a function of age and race. A multifaceted approach is used to quantify these relationships. A causal study is performed to explore the effect of various comorbidities on patient outcomes. Least squares regression models are developed to evaluate and compare significant factors that influence total charges and length of stay. Logistic regression is used to study the factors that may cause patient mortality or transferring. In addition, different survival models are developed to study the impact of comorbidity on length of stay with censoring information. This study shows the interactions and relationship among various comorbidities and breast cancer. It shows that certain hypertension may not increase length of stay and total charges; diabetes behaves differently among general population and breast cancer patients; mental disorder has an impact on patient disposition that affects true length of stay and charges, and obesity may have limited effect on patient outcomes. Moreover, this study will help to better understand the expenditure patterns for population subgroups with several chronic conditions and to quantify the impact of comorbidities on patient outcomes. Lastly, it also provides insight for breast cancer patients with comorbidities as a function of age and race.     

Estimating the benefits of mammography screening: the impact of study design

BACKGROUND: Mammography screening is justifiable only if it leads to reduction in breast cancer mortality. However, evaluation of routine screening is not straightforward, as no unscreened control group is available. We report here on a cohort study of the effect of routine mammography on breast cancer mortality, and illustrate how variations in the analytic approach can affect the conclusions. METHODS: We used data from the mammography screening program in Copenhagen, Denmark, for the period 1991-2001. We used local historical, concurrent regional, and historical regional control groups, and included only deaths from breast cancers diagnosed during the observation periods. We examined the impact of various control groups, of including all breast cancer deaths, and of using individual data versus routine statistics. RESULTS: Combining all 3 control groups gave an estimated 25% reduction in breast cancer mortality. The estimate was 20% using only a local historical control group, and 9% using only a concurrent regional control group. Including all breast cancer deaths resulted in an estimate of 21% reduction in breast cancer mortality. Using routine statistics and a concurrent regional control group resulted in an estimated increase of 6% in breast cancer mortality. CONCLUSION: Estimated changes in breast cancer mortality following the introduction of routine mammography ranged from a 25% reduction (based on the best methodology) to a 6% increase with a less rigid study design. The estimated effect of routine mammography on breast cancer mortality is thus highly dependent on study design.     

微量元素与乳腺癌死亡率的模式识别研究

采用基于统计学习理论的支持向量机方法,建立了硒、铜、锌、镉、铬、锰和砷7种微量元素的平均摄入量与每10万人中的乳腺癌死亡人数的预测模型,并对6个国家或地区进行了预测,还与传统的人工神经网络方法进行了比较。结果表明:在根据微量元素摄入量进行乳腺癌发病率预测方面,支持向量机方法有明显的优势。     

Breast cancer and women's labor supply

OBJECTIVE: To investigate the effect of breast cancer on women's labor supply. DATE SOURCE/STUDY SETTING: Using the 1992 Health and Retirement Study, we estimate the probability of working using probit regression and then, for women who are employed, we estimate regressions for average weekly hours worked using ordinary least squares (OLS). We control for health status by using responses to perceived health status and comorbidities. For a sample of married women, we control for spouses' employer-based health insurance. We also perform additional analyses to detect selection bias in our sample. PRINCIPAL FINDINGS: We find that the probability of breast cancer survivors working is 10 percentage points less than that for women without breast cancer. Among women who work, breast cancer survivors work approximately three more hours per week than women who do not have cancer. Results of similar magnitude persist after health status is controlled in the analysis, and although we could not definitively rule out selection bias, we could not find evidence that our results are attributable to selection bias. CONCLUSIONS: For some women, breast cancer may impose an economic hardship because it causes them to leave theirjobs. However, for women who survive and remain working, this study failed to show a negative effect on hours worked associated with breast cancer. Perhaps the morbidity associated with certain types and stages of breast cancer and its treatment does not interfere with work.     

Predictors of Medicare costs in elderly beneficiaries with breast, colorectal, lung, or prostate cancer

Background: Determining the apportionment of costs of cancer care and identifying factors that predict costs are important for planning ethical resource allocation for cancer care, especially in markets where managed care has grown. Design: This study linked tumor registry data with Medicare administrative claims to determine the costs of care for breast, colorectal, lung and prostate cancers during the initial year subsequent to diagnosis, and to develop models to identify factors predicting costs. Subjects: Patients with a diagnosis of breast (n=1,952), colorectal (n=2,563), lung (n=3,331) or prostate cancer (n=3,179) diagnosed from 1985 through 1988. Results: The average costs during the initial treatment period were $12,141 (s.d.=$10,434) for breast cancer, $24,910 (s.d.=$14,870) for colorectal cancer, $21,351 (s.d.=$14,813) for lung cancer, and $14,361 (s.d.=$11,216) for prostate cancer. Using least squares regression analysis, factors significantly associated with cost included comorbidity, hospital length of stay, type of therapy, and ZIP level income for all four cancer sites. Access to health care resources was variably associated with costs of care. Total R ² ranged from 38% (prostate) to 49% (breast). The prediction error for the regression models ranged from <1% to 4%, by cancer site. Conclusions: Linking administrative claims with state tumor registry data can accurately predict costs of cancer care during the first year subsequent to diagnosis for cancer patients. Regression models using both data sources may be useful to health plans and providers and in determining appropriate prospective reimbursement for cancer, particularly with increasing HMO penetration and decreased ability to capture complete and accurate utilization and cost data on this population. This revised version was published online in July 2006 with corrections to the Cover Date.     

Regression analysis with multiplicative and time-varying additive regression coefficients with examples from breast and colon cancer

Regression analysis may be used to simplify the representation of mortality rates when there are many significant prognostic covariates or to adjust for confounding effects. The principal request of the regression model in this range of use is to have unbiased parameter estimates. A model with constant multiplicative and time-varying additive regression coefficients is discussed. The model allows some covariate effects to be multiplicative while allowing others to have a time-varying additive effect. Thus, it is a mix of classical Cox regression and Aalen's additive risk model. A major characteristic of cancer mortality rates, in contrast to general mortality rates, is that hazard rates, after a potentially initial increase, decrease, although not always tending to zero. Cancer diseases, like breast and colon cancer, have significantly increased cause-specific mortality rates even 20 years after diagnosis. Another major feature in cancer survival analysis is that many covariate effects are time-varying. Some covariate effects, like age at diagnosis, may only be significant for a limited time after diagnosis. Furthermore, some treatment procedures may initially decrease the mortality, while the long-term effect may be opposite. A third issue is that average covariate effects are very often not multiplicative. Estimation is carried out iteratively; the cumulative additive regression functions are estimated non-parametrically using a least-squares method and the multiplicative parameters are estimated from the partial likelihood. The method is applied on 3201 female breast cancer and 1372 male colon cancer patients.