Trained as a systems engineer and quantitative modeler, my research works focus on the development and applications of data analytics, information technology, and systems science approach to mental and behavioral health issues.
Before joining in the USC School of Social Work in 2016 as a postdoc, I was a doctoral student in Industrial and Systems Engineering. My dissertation works involved a series of three papers that developed and applied predictive analytics to forecasting depression and improving health services for low-income, predominantly older minority patients in primary care. These papers generated evidence of the predictive accuracy, and costs and benefits from using the prediction models to facilitate more patient-centered depression care. A proposal for this series of studies was picked as the winner from among 32 competitors in Pathways to Clinical Forecasting Pilot Grant Competition organized by UCLA. A paper of this work has led me to be awarded with Honorable Mention in the 2015 Student Paper Competition (among 59 submissions) along with a publication in the Preventing Chronic Disease journal, an official journal of the U.S. Centers for Disease Control and Prevention (CDC).
Since appointed as Research Assistant Professor in 2018, I am growing my research to explore new potentials of quantitative modeling in mental and behavioral health situations. I am developing a new series of research that uses computational techniques, such as simulation, Markov modeling, and reinforcement learning, to understand and mitigate the negative impacts of affective disorders on workplace decision making and performance. I also join in new collaborations to explore applications of quantitative modeling techniques in studying various behaviors such as substance use and patient self-care.
Besides my research work, I am a faculty member of the Data and Statistical Core Team at the Hamovitch Research Center. I am providing statistical and quantitative modeling consulting to faculties and students across the school.
University of Southern California: Ph.D., Industrial and Systems Engineering 2016
Dissertation: Clinical prediction models to forecast depression in patients with diabetes and applications in depression screening policymaking
University of Southern California: M.S., Operations Research 2015
Zhejiang University: M.S., Industrial Engineering 2012
Zhejiang University: B.S., Industrial Engineering 2009
Areas of Expertise (5)
Honorable Mention, 2015 Annual Student Research Paper Contest of the Preventing Chronic Disease journal (professional)
Honorable Mention for the paper "Development of a clinical forecasting model to predict comorbid depression among diabetes patients and an application in depression screening policy making". Preventing Chronic Disease is an official journal of the U.S. Centers for Disease Prevention and Control.
Honorable Mention, 2014 Annual Student Research Paper Contest of the Preventing Chronic Disease journal (professional)
Honorable Mention for the paper "Collaborative depression care among Latino patients in diabetes disease management, Los Angeles, 2011-2013". Preventing Chronic Disease is an official journal of the U.S. Centers for Disease Prevention and Control.
Winner, Pathways to Clinical Forecasting Pilot Grant Competition, 2014 (professional)
The wining proposal titled "Predicting Depression Outcomes to Facilitate Large-Scale Depression Management".
Best Poster Award, 2nd International Conference in Big Data and Analytics in Healthcare, Singapore, 2014 (professional)
Awarded for the poster "Developing depression symptoms prediction models to improve depression care outcomes: Preliminary results".
Finalist, Lee B. Lusted Student Prizes, Society for Medical Decision Making (SMDM), 2015 (professional)
Competition research title: "Development of a Clinical Forecasting Model for Detecting Comorbid Depression among Patients with Diabetes and an Application in Depression Screening Policymaking".
Research Reports & Projects (1)
Latinos & Alzheimer’s disease, new numbers behind the crisis
Projection of the costs for U.S. Latinos living with Alzheimer’s Disease through 2060
U.S. Latinos living with Alzheimer’s disease are projected to increase from 379,000 in 2012 to 1.1 million by 2030 and to 3.5 million by 2060—a growth of 832 percent. In addition, the cumulative direct and indirect costs of Alzheimer’s disease on the U.S. Latino community, including millions of family caregivers, will ultimately cost the U.S. economy $373 billion by 2030 and $2.35 trillion (in 2012 dollars) by 2060.
Articles & Publications (6)
Wu S, Ell K, Jin H, Vidyanti I, Chou CP, etc.
Comorbid depression is a significant challenge for safety-net primary care systems. Team-based collaborative depression care is effective, but complex system factors in safety-net organizations impede adoption and result in persistent disparities in outcomes. The aim of this study was to compare 6-month outcomes in a large implementation study of a technology-facilitated care model with a usual care model and a supported care model that involved team-based collaborative depression care for safety-net primary care adult patients with type 2 diabetes. Using a comparative effectiveness design, the study enrolled 1406 patients (484 in usual care, 480 in supported care, and 442 in technology-facilitated care), most of whom were Hispanic or Latino and female. A generalized propensity score analysis showed that comparing with usual care, both the supported care and technology-facilitated care groups were associated with significant reduction in depressive symptoms measured by scores on the 9-item Patient Health Questionnaire, decreased prevalence of major depression and reduced functional disability as measured by Sheehan Disability Scale scores. Technology-facilitated care was significantly associated with depression remission, increased satisfaction with care for emotional problems among depressed patients, reduced total cholesterol level, improved satisfaction with diabetes care, and increased odds of taking an glycated hemoglobin test. The study revealed that Both the technology-facilitated care and supported care delivery models showed potential to improve 6-month depression and functional disability outcomes. The technology-facilitated care model has a greater likelihood to improve depression remission, patient satisfaction, and diabetes care quality.
Hay J, Lee PJ, Jin H, Guterman J, Gross-Schulman S, Ell K, Wu S
BACKGROUND: The Diabetes-Depression Care-Management Adoption Trial is a translational study of safety-net primary care predominantly Hispanic/Latino patients with type 2 diabetes in collaboration with the Los Angeles County Department of Health Services.
OBJECTIVES: To evaluate the cost-effectiveness of an information and communication technology (ICT)-facilitated depression care management program.
METHODS: Cost-effectiveness of the ICT-facilitated care (TC) delivery model was evaluated relative to a usual care (UC) and a supported care (SC) model. TC added automated low-intensity periodic depression assessment calls to patients. Patient-reported outcomes included the 12-Item Short Form Health Survey converted into quality-adjusted life-years (QALYs) and the 9-Item Patient Health Questionnaire-calculated depression-free days (DFDs). Costs and outcomes data were collected over a 24-month period (-6 to 0 months baseline, 0 to 18 months study intervention).
RESULTS: A sample of 1406 patients (484 in UC, 480 in SC, and 442 in TC) was enrolled in the nonrandomized trial. TC had a significant improvement in DFDs (17.3; P = 0.011) and significantly greater 12-Item Short Form Health Survey utility improvement (2.1%; P = 0.031) compared with UC. Medical costs were statistically significantly lower for TC (-$2328; P = 0.001) relative to UC but not significantly lower than for SC. TC had more than a 50% probability of being cost-effective relative to SC at willingness-to-pay thresholds of more than $50,000/QALY.
CONCLUSIONS: An ICT-facilitated depression care (TC) delivery model improved QALYs, DFDs, and medical costs. It was cost-effective compared with SC and dominant compared with UC.
Ramirez M, Wu S, Jin H, Ell K, Schulman SG, Sklaroff L, Guterman J
Remote patient monitoring is increasingly integrated into health care delivery to expand access and increase effectiveness. Automation can add efficiency to remote monitoring, but patient acceptance of automated tools is critical for success. From 2010 to 2013, the Diabetes-Depression Care-management Adoption Trial (DCAT)–a quasi-experimental comparative effectiveness research trial aimed at accelerating the adoption of collaborative depression care in a safety-net health care system–tested a fully automated telephonic assessment (ATA) depression monitoring system serving low-income patients with diabetes. The aim of this study was to determine patient acceptance of ATA calls over time, and to identify factors predicting long-term patient acceptance of ATA calls. The study found that at 6 and 12 months, respectively, 89.6% (69/77) and 63.7% (49/77) of patients “agreed” or “strongly agreed” that they would be willing to use ATA calls in the future. At 18 months, 51.0% (64/125) of patients perceived ATA calls as useful and 59.7% (46/77) were willing to use the technology. Moreover, in the first 6 months, most patients reported that ATA calls felt private/secure (75.9%, 82/108) and were easy to use (86.2%, 94/109), useful (65.1%, 71/109), and nonintrusive (87.2%, 95/109). Perceived usefulness, however, decreased to 54.1% (59/109) in the second 6 months of the trial. Factors predicting willingness to use ATA calls at the 18-month follow-up were perceived privacy/security and long-term perceived usefulness of ATA calls. No patient characteristics were significant predictors of long-term acceptance. As a conclusion, in the short term, patients are generally accepting of ATA calls for depression monitoring, with ATA call design and the care management intervention being primary factors influencing patient acceptance. Acceptance over the long term requires that the system be perceived as private/secure, and that it be constantly useful for patients’ needs of awareness of feelings, self-care reminders, and connectivity with health care providers.
Jin H, Wu S, Di Capua P
Introduction: Depression is a common but often undiagnosed comorbid condition of people with diabetes. Mass screening can detect undiagnosed depression but may require significant resources and time. The objectives of this study were 1) to develop a clinical forecasting model that predicts comorbid depression among patients with diabetes and 2) to evaluate a model-based screening policy that saves resources and time by screening only patients considered as depressed by the clinical forecasting model.
Methods: We trained and validated 4 machine learning models by using data from 2 safety-net clinical trials; we chose the one with the best overall predictive ability as the ultimate model. We compared model-based policy with alternative policies, including mass screening and partial screening, on the basis of depression history or diabetes severity.
Results: Logistic regression had the best overall predictive ability of the 4 models evaluated and was chosen as the ultimate forecasting model. Compared with mass screening, the model-based policy can save approximately 50% to 60% of provider resources and time but will miss identifying about 30% of patients with depression. Partial-screening policy based on depression history alone found only a low rate of depression. Two other heuristic-based partial screening policies identified depression at rates similar to those of the model-based policy but cost more in resources and time.
Conclusion: The depression prediction model developed in this study has compelling predictive ability. By adopting the model-based depression screening policy, health care providers can use their resources and time better and increase their efficiency in managing their patients with depression.
Jin H, Wu S, Vidyanti I, Di Capua P, Wu B
BACKGROUND: Depression is a common and often undiagnosed condition for patients with diabetes. It is also a condition that significantly impacts healthcare outcomes, use, and cost as well as elevating suicide risk. Therefore, a model to predict depression among diabetes patients is a promising and valuable tool for providers to proactively assess depressive symptoms and identify those with depression.
OBJECTIVES: This study seeks to develop a generalized multilevel regression model, using a longitudinal data set from a recent large-scale clinical trial, to predict depression severity and presence of major depression among patients with diabetes.
METHODS: Severity of depression was measured by the Patient Health Questionnaire PHQ-9 score. Predictors were selected from 29 candidate factors to develop a 2-level Poisson regression model that can make population-average predictions for all patients and subject-specific predictions for individual patients with historical records. Newly obtained patient records can be incorporated with historical records to update the prediction model. Root-mean-square errors (RMSE) were used to evaluate predictive accuracy of PHQ-9 scores. The study also evaluated the classification ability of using the predicted PHQ-9 scores to classify patients as having major depression.
RESULTS: Two time-invariant and 10 time-varying predictors were selected for the model. Incorporating historical records and using them to update the model may improve both predictive accuracy of PHQ-9 scores and classification ability of the predicted scores. Subject-specific predictions (for individual patients with historical records) achieved RMSE about 4 and areas under the receiver operating characteristic (ROC) curve about 0.9 and are better than population-average predictions.
CONCLUSIONS: The study developed a generalized multilevel regression model to predict depression and demonstrated that using generalized multilevel regression based on longitudinal patient records can achieve high predictive ability.
Wu B, Jin H, Vidyanti I, Lee PJ, Ell K, Wu S
INTRODUCTION: The prevalence of comorbid diabetes and depression is high, especially in low-income Hispanic or Latino patients. The complex mix of factors in safety-net care systems impedes the adoption of evidence-based collaborative depression care and results in persistent disparities in depression outcomes. The Diabetes-Depression Care-Management Adoption Trial examined whether the collaborative depression care model is an effective approach in safety-net clinics to improve clinical care outcomes of depression and diabetes.
METHODS: A sample of 964 patients with diabetes from 5 safety-net clinics were enrolled in a quasi-experimental study that included 2 arms: usual care, in which primary medical providers and staff translated and adopted evidence-based depression care; and supportive care, in which providers of a disease management program delivered protocol-driven depression care. Because the study design established individual treatment centers as separate arms, we calculated propensity scores that interpreted the probability of treatment assignment conditional on observed baseline characteristics. Primary outcomes were 5 depression care outcomes and 7 diabetes care measures. Regression models with propensity score covariate adjustment were applied to analyze 6-month outcomes.
RESULTS: Compared with usual care, supportive care significantly decreased Patient Health Questionnaire-9 scores, reduced the number of patients with moderate or severe depression, improved depression remission, increased satisfaction in care for patients with emotional problems, and significantly reduced functional impairment.
CONCLUSION: Implementing collaborative depression care in a diabetes disease management program is a scalable approach to improve depression outcomes and patient care satisfaction among patients with diabetes in a safety-net care system.