The Biomedical Statistics Data Science Lab Team Members

The Biomedical Statistics & Data Science Lab Team Members: • Yaming Li, MD, MS. Lead Statistician for Clinical Research • Avantika Srivastava, MS. Lead Statistician for causal inference and biomarker models • Dan Lavage. Lead Statistician for intervention studies Lab Director: • Doug Landsittel, Ph. D. • dpl 12@pitt. edu Website: landsittellab. pitt. edu • Kristin Kropf, MS. Instructional Developer and Program Coordinator Lauren Rost and Smitha Edakalavan, MS. National Libraries of Medicine Ph. D Trainees Barbara Karnbauer. Administrative Director Professor, Biomedical Informatics, Biostatistics, Medicine, and Clinical and Translational Science Director, Biomedical Statistics & Data Science Lab & Director of Biostatistics (Research) Starzl Transplant Institute Program Director, Expanding National Capacity in PCOR through Training & Collaboration (ENACT) Network Director, Data Coordinating and Image Analysis Center, Consortium for Radiological Imaging Studies of PKD Director, Data Coordinating Center, Society of Critical Care Medicine Discovery Network Chair, Safety and Occupational Health Study Section, CDC/NIOSH Member, Comparative Effectiveness Research Center, Institute for Clinical Research Education, Center for Research on Health

The Consortium for Radiological Imaging Studies of PKD (CRISP) Total kidney volume (TKV) increases exponentially in autosomal dominant polycystic kidney disease (ADPKD). TKV growth predicts faster decline in renal function. Challenge: Measurement of renal function is highly variable and does not decline steadily until it is too late to prevent end-stage disease. Impact: ADPKD patients with steeper increases in TKV can be identified years before progressive decline in GFR. Publications: 14 published manuscrips (2012 -). • Yu, et al. (2019) Long-term trajectory of kidney function in autosomal dominant polycystic kidney disease. Kidney Int. • Bae, et al. (2019) Growth Pattern of Kidney Cyst Number and Volume in Autosomal Dominant Polycystic Kidney Disease. Clin. J. of the Amer. Soc. of Nephrology. • Bae, et al. (In Press) Expanded Imaging Classification of Autosomal Dominant Polycystic Kidney Disease. J. of the Amer. Soc. of Nephrology. Funding: NIDDK 1 R 01 DK 113111 -01

A Decision Tool for Causal Inference and Observational Data Analysis Methods (DECODE CER) DECODE CER is a PCORI-funded tool in Challenge: Causal inference can be expressed in terms of potential outcomes. Time 0 Google Drive with links to key resources. effects cannot be directly estimated with standard observational methods. Educational resources are critically needed for improving the science of clinical decision-making Impact: We have developed a comprehensive open educational resource for complex methods and rigorous designs. Treatment Outcome at time 1 A Yi(A) B Yi(B) Yi(A) and Yi(B) are potential outcomes, where one is observed, and one is counterfactual. The Causal Effect = expected difference of Yi(A) - Yi(B). Publication: Landsittel D, et al. (2019) Guidance for Researchers on Optimal Methods for Conducting Comparative Effectiveness Research With Observational Data. Washington, DC: Patient-Centered Outcomes Research Institute (PCORI). Funding: PCORI R-IMC-1306 -03827

The Expanding National Capacity in PCOR through Training & Collaboration (ENACT) Program Collaboration with 7 (including 6 Minority-Serving Institutions) to build expertise in patient-centered outcomes research. Publication: Landsittel et al. (2017) Training in Patient. Centered Outcomes Research for Specific Researcher Communities. J Clin Transl Sci. Funding: AHRQ 1 R 25 HS 023185 -01 Challenge: Methods for patient-centered outcomes research (PCOR) are complex, but critical for optimal clinical decisions. Impact: • ENACT has trained 22 Fellows in PCOR methods. • Over 100 other trainees participated in online courses on writing a successful concept proposal. • ENACT produced an online course, which is publicly-available through Google Drive at https: //www. landsittellab. pitt. edu/educationalresources. • The online resource is also being used to facilitate a flipped classroom for the University of Pittsburgh PCOR course (CLRES 2107). • The ENACT Network is continuing collaboration on multiple projects and facilitating further networking between the University of Pittsburgh, Minority-Serving Institutions, and other institutions focused on PCOR and health disparities research.

Statistical Methods & Educational Resources at the Intersection of Predictive Models & Intervention Research The Big Topics Biomarkers & Prediction Intervention Research Causal Inference Outcomes in COVID-19. Machine Learning Methods. Personalized Medicine Propensity score-based methods Prognosis of end-stage kidney disease & graft failure The Big Clinical Studies Objectives Collaborative Research Basic Science Quasi-experimental Designs Immunotherapy Trials Comparative Effectiveness Research More accurate prediction Guide Methods Development Unbiased Causal Inference More rigorous designs Educational Resources to Improve Science