New York City Metropolitan Area
support lung, esophageal and biliary tract cancer with filing experience
support lung, esophageal and biliary tract cancer with experiences in study designs, clinical data reporting and CSRs
1. A novel graphical approach for sequentially rejective multiple testing procedures is developed, which can reject more than one hypothesis at each testing step. The proposed graphical approach is more efficient and flexible than the existing graphical approaches. Theoretically, we prove that the proposed graphical approach strongly controls the FWER under arbitrary dependence by employing the proposed extended sequential rejection principle, which provides an alternative way of proving the FWER control. 2. A new graphical approach for logically restricted multiple hypotheses tests using re-assignment mechanism is developed. The proposed approach distinguishes the clinical importance and logical conditions, and make full use of the critical values from logically restricted hypotheses. It is more intuitive and easier to perform multiple hypotheses with complex structures than the few existing methods, such as the entangled graphical approach. 3. Develop powerful selective inference methods in large scale multiple hypothesis testing. Three different solutions are provided to improve existing two-stage multiple testing procedures (TS-MTPs), which are adaptive two-stage Bonferroni procedure, selective parallel gatekeeping procedure and data-driven weighted selective procedure. Extensive simulation studies show the better performance in sense of average power of the proposed three procedures than existing TS-MTPs in some certain scenarios.
Leading recitations of undergraduate math courses, grading assignment and quizzes.
Teaching an undergraduate-level statistics course -Statistics for Engineering . Assessing their performance by making and grading their homework, quizzes, midterm and final exams.
Customer Feedback Analysis 1. In order to leverage customer feedback to better understand customer satisfaction, retention and prioritize investment opportunities, advanced text analytics approach is used for topic modeling 2. Based on created topics, a predictive model is built to predict net promoter score (NPS) by using logistic regression method. 3. Correspondence analysis is conducted to link the association between created topics and the demographics among Healthfirst members. 4. Issue prioritization analysis is developed, such that the topics with lower performance and higher impact on Healthfirst business are main issue topics. It provides a focus area for further marketing and investment strategies making. Opioid Crisis Investigation among Healthfirst Medicaid Members 1. From amazon web services (AWS) redshift, data including member claims, pharmacy and service lines are pulled and joined together by SQL. 2. Based on four main metrics: demographics, pharmacy shopping, diagnosis history and utilization history, a predictive model of opioid dependence is built by using logistics regression method. 3. Live risk scores for each Medicaid member are calculated in Tableau, such that health care manager is able to take further actions for members with high risk scores.