Yinfei Kong

Innovated interaction screening for high-dimensional nonlinear classification

This paper is concerned with the problems of interaction screening and nonlinear classification in a high-dimensional setting. We propose a two-step procedure, IIS-SQDA, where in the first step an innovated interaction screening (ITS) approach based on transforming the original p-dimensional feature vector is proposed, and in the second step a sparse quadratic discriminant analysis (SQDA) is proposed for further selecting important interactions and main effects and simultaneously conducting classification. Our IIS approach screens important interactions by examining only p features instead of all two-way interactions of order O(p(2)). Our theory shows that the proposed method enjoys sure screening property in interaction selection in the high-dimensional setting of p growing exponentially with the sample size. In the selection and classification step, we establish a sparse inequality on the estimated coefficient vector for QDA and prove that the classification error of our procedure can be upper-bounded by the oracle classification error plus some smaller order term. Extensive simulation studies and real data analysis show that our proposal compares favorably with existing methods in interaction selection and high-dimensional classification.

Interaction pursuit in high-dimensional multi-response regression via distance correlation

Feature interactions can contribute to a large proportion of variation in many prediction models. In the era of big data, the coexistence of high dimensionality in both responses and covariates poses unprecedented challenges in identifying important interactions. In this paper, we suggest a two-stage interaction identification method, called the interaction pursuit via distance correlation (IPDC), in the setting of high-dimensional multi-response interaction models that exploits feature screening applied to transformed variables with distance correlation followed by feature selection. Such a procedure is computationally efficient, generally applicable beyond the heredity assumption, and effective even when the number of responses diverges with the sample size. Under mild regularity conditions, we show that this method enjoys nice theoretical properties including the sure screening property, support union recovery, and oracle inequalities in prediction and estimation for both interactions and main effects. The advantages of our method are supported by several simulation studies and real data analysis.

Improving Coordination of Addiction Health Services Organizations with Mental Health and Public Health Services

In this mixed-method study, we examined coordination of mental health and public health services in addiction health services (AHS) in low-income racial and ethnic minority communities in 2011 and 2013. Data from surveys and semistructured interviews were used to evaluate the extent to which environmental and organizational characteristics influenced the likelihood of high coordination with mental health and public health providers among outpatient AHS programs. Coordination was defined and measured as the frequency of interorganizational contact among AHS programs and mental health and public health providers. The analytic sample consisted of 112 programs at time 1 (T1) and 122 programs at time 2 (T2), with 61 programs included in both periods of data collection. Forty-three percent of AHS programs reported high frequency of coordination with mental health providers at T1 compared to 66% at T2. Thirty-one percent of programs reported high frequency of coordination with public health services at T1 compared with 54% at T2. Programs with culturally responsive resources and community linkages were more likely to report high coordination with both services. Qualitative analysis highlighted the role of leadership in leveraging funding and developing creative solutions to deliver coordinated care. Overall, our findings suggest that AHS program funding, leadership, and cultural competence may be important drivers of program capacity to improve coordination with health service providers to serve minorities in an era of health care reform.

Paths to improving engagement among racial and ethnic minorities in addiction health services

BackgroundMembers of racial and ethnic minority groups are most likely to experience limited access and poor engagement in addiction treatment. Research has been limited on the role of program capacity and delivery of comprehensive care in improving access and retention among minorities with drug abuse issues. The goal of this study was to examine the extent to which access and retention are enhanced when racial and ethnic minorities receive care from high-capacity addiction health services (AHS) programs and via coordination with mental health and receipt of HIV testing services.MethodsThis multilevel cross-sectional analysis involved data from 108 programs merged with client data from 2011 for 13,478 adults entering AHS. Multilevel negative binomial regression models were used to test interactions and indirect relationships between program capacity and days to enter treatment (wait time) and days in treatment (retention).ResultsCompared to low-capacity programs and non-Latino and non-African American clients, Latinos and African Americans served in high-capacity programs reported shorter wait times to admission, as hypothesized. African Americans also had longer treatment retention in high-capacity programs. Receipt of HIV testing and program coordination of mental health services played an indirect role in the relationship between program capacity and wait time.ConclusionsProgram capacity and coordinated services in AHS may reduce disparities in access to care. Implications for supporting low-capacity programs to eliminate the disparity gap in access to care are discussed.

Advancing theory development: exploring the leadership–climate relationship as a mechanism of the implementation of cultural competence

BackgroundLeadership style and specific organizational climates have emerged as critical mechanisms to implement targeted practices in organizations. Drawing from relevant theories, we propose that climate for implementation of cultural competence reflects how transformational leadership may enhance the organizational implementation of culturally responsive practices in health care organizations.MethodsUsing multilevel data from 427 employees embedded in 112 addiction treatment programs collected in 2013, confirmatory factor analysis showed adequate fit statistics for our measure of climate for implementation of cultural competence (Cronbach’s alpha = .88) and three outcomes: knowledge (Cronbach’s alpha = .88), services (Cronbach’s alpha = .86), and personnel (Cronbach’s alpha = .86) practices.ResultsResults from multilevel path analyses indicate a positive relationship between employee perceptions of transformational leadership and climate for implementation of cultural competence (standardized indirect effect = .057, bootstrap p < .001). We also found a positive indirect effect between transformational leadership and each of the culturally competent practices: knowledge (standardized indirect effect = .006, bootstrap p = .004), services (standardized indirect effect = .019, bootstrap p < .001), and personnel (standardized indirect effect = .014, bootstrap p = .005).ConclusionsFindings contribute to implementation science. They build on leadership theory and offer evidence of the mediating role of climate in the implementation of cultural competence in addiction health service organizations.

Barriers to accessing substance abuse treatment in Mexico: national comparative analysis by migration status

BackgroundWe examined Mexican migrants’ perceived barriers to entering substance abuse treatment and potential differences by gender.MethodsThis study analyzed a subset of household data collected in Mexico in 2011 via the Encuesta Nacional de Adicciones (National Survey of Addictions). A sample of 1,143 individuals who reported using illicit drugs was analyzed using multivariate negative binomial models to determine direct and moderated relationships of gender, migrant status, and drug dependence with perceived barriers to accessing treatment.ResultsSignificant findings included disparities in drug dependence by migrant status. Compared with non-migrant men, women who have traveled to the United States was associated with fewer (1.3) barriers to access treatment. Fewer barriers to access care were associated with individuals residing in other regions of the country, compared to those living in Mexico City.ConclusionsDrug dependence, gender, migration status and regional location are factors associated with access to needed treatment. Implications for health care policy to develop treatment services infrastructure and for future research are discussed in the context of ongoing drug policy reform in Mexico.

Scalable Inference for Massive Data

Abstract With the availability of massive data sets, how to make accurate inference with lower computational cost is the key to improving scalability. One important scenario is when both the sample size and the number of covariates are large, which is in contrast to the typical high-dimensional setting with relatively low sample size. In such cases, naive application of existing sparse modeling procedures can be computationally inefficient or infeasible. To ameliorate the scalability, in this paper we suggest the method of inference with partitioned data (IPAD) that divides the entire sample set into subsamples for correcting the bias and constructs confidence intervals by aggregating the estimates based on subsamples. Compared to inference with the whole sample set, such an approach can substantially reduce the computational cost. Furthermore, we establish confidence intervals of the bagging estimator for aggregation, which remain largely unexplored in the literature due to the communication barriers between subsamples. Both computational advantage and theoretical guarantee of our new method are evidenced by numerical examples.

Achieving Sobriety Among Latino Older Adults

Research on culturally responsive interventions for substance use disorders among Latino older adults has been limited. This study examined the role of counselors’ Spanish language proficiency, among other indicators of quality of care, and client factors, such as gender and drug of choice, in the odds of sobriety at treatment discharge among Latino older adults. Treatment retention was associated with higher odds of sobriety, as was female gender. Participants who reported using cocaine or crack were twice as likely to be sober at discharge as those using heroin. Counselors’ Spanish language proficiency was associated with lower odds of client sobriety, an unexpected finding. Implications for research and policy are discussed, particularly the need to explore how these client and organizational characteristics affect the recovery of Latino older adults with substance use problems.