BEGIN:VCALENDAR PRODID:-//github.com/ical-org/ical.net//NONSGML ical.net 4.0//EN VERSION:2.0 BEGIN:VTIMEZONE TZID:America/New_York X-LIC-LOCATION:America/New_York BEGIN:STANDARD DTSTART:20241103T020000 RRULE:FREQ=YEARLY;BYDAY=1SU;BYMONTH=11 TZNAME:EST TZOFFSETFROM:-0400 TZOFFSETTO:-0500 END:STANDARD BEGIN:DAYLIGHT DTSTART:20250309T020000 RRULE:FREQ=YEARLY;BYDAY=2SU;BYMONTH=3 TZNAME:EDT TZOFFSETFROM:-0500 TZOFFSETTO:-0400 END:DAYLIGHT END:VTIMEZONE BEGIN:VEVENT DESCRIPTION:Note: BIS 526 students are required to attend in person. Other s are invited to attend in person but may also attend via zoom. Speaker- Rebecca Betensky\, PhD Title- “Two problems of estimation in the presence of semi competing or competing risks” Abstract I will discuss two proble ms that involve competing risks. The first problem is that of estimation in the presence of semicompeting risks and left truncation. I will presen t three nonparametric maximum likelihood estimators under different model s for censoring. The second problem is that of estimation in the presence of a competing risk with an intermediate event that terminates the risk of a different competing risk. In the absence of censoring\, this early t ermination of risk can be ignored for estimation of the cumulative incide nce function (CIF). In the presence of censoring\, I will show that in la rge samples\, early termination of risk can be ignored as well for certai n CIF estimators. I will discuss the implications of this for regression modeling via the cause specific hazard and via the subdistribution hazard . I demonstrate these results in simulations and an analysis of dementia with recovered COVID\, with the competing risk of dementia with Long COVI D and intermediate terminating event of onset of Long COVID.\n\nAdmission :\nFree\n\nDetails URL:\nhttps://medicine.yale.edu/event/ysph-biostatisti cs-seminar-1/\n DTEND;TZID=America/New_York:20260113T130000 DTSTAMP:20260517T040622Z DTSTART;TZID=America/New_York:20260113T120000 GEO:41.302961;-72.931638 LOCATION:106 A&B\, 47 College Street\, New Haven\, CT\, United States RECURRENCE-ID;TZID=America/New_York:20260113T120000 SEQUENCE:0 STATUS:Confirmed SUMMARY:YSPH Biostatistics Seminar- “Two problems of estimation in the pre sence of semi competing or competing risks” UID:f0a07cbc-40b8-4d98-ab72-aac4972c18b7 END:VEVENT BEGIN:VEVENT DESCRIPTION:Note: BIS 526 students are required to attend in person. Other s are invited to attend in person but may also attend via zoom. Speaker D r. Rebecca A. Hubbard Title: “The Brave New World of Real-World Data” Abs tract Real-world data (RWD) are ubiquitous\, creating exciting opportunit ies to advance science and society through data-driven decision-making. H owever\, data quality and availability vary across populations and settin gs\, creating significant risks of bias. In this talk\, I will highlight examples where evidence from clinical trials and epidemiological cohorts can be augmented with RWD to enhance both the timeliness and generalizabi lity of findings\, and I will describe statistical and design-based strat egies to harness the potential of RWD while safeguarding reliability. I w ill discuss approaches for extending evidence from randomized trials to t rial-ineligible populations\, hybrid controlled trial designs that integr ate RCT and RWD components\, and methods that combine RWD with deeply cha racterized epidemiological cohorts to address data quality limitations. I llustrations from cancer and aging research will highlight the opportunit ies and pitfalls of this evolving evidence ecosystem and emphasize that t he contribution of RWD to evidence-based decision making rests on the rig or of the methods with which they are analyzed.\n\nAdmission:\nFree\n\nDe tails URL:\nhttps://medicine.yale.edu/event/ysph-biostatistics-seminar-2/ \n DTEND;TZID=America/New_York:20260120T130000 DTSTAMP:20260517T040622Z DTSTART;TZID=America/New_York:20260120T120000 GEO:41.302961;-72.931638 LOCATION:106 A&B\, 47 College Street\, New Haven\, CT\, United States RECURRENCE-ID;TZID=America/New_York:20260120T120000 SEQUENCE:0 STATUS:Confirmed SUMMARY:YSPH Biostatistics Seminar- “The Brave New World of Real-World Dat a” UID:f0a07cbc-40b8-4d98-ab72-aac4972c18b7 END:VEVENT BEGIN:VEVENT DESCRIPTION:Note: BIS 526 students are required to attend in person. Other s are invited to attend in person but may also attend via zoom. Speaker D r. Pei Wang Title “Learning directed acyclic graphs for ligands and recep tors based on spatially resolved transcriptomic data of ovarian cancer” A bstract To unravel the mechanism of immune activation and suppression wit hin tumors\, a critical step is to identify transcriptional signals gover ning cell-cell communication between tumor and immune/stromal cells in th e tumor microenvironment. Central to this communication are interactions between secreted ligands and cell-surface receptors\, creating a highly c onnected signaling network among cells. Recent advancements in in situ-om ics profiling\, particularly spatial transcriptomic (ST) technology\, pro vide unique opportunities to directly characterize ligand-receptor signal ing networks that power cell-cell communication. In this paper\, we propo se a novel statistical method\, LRnetST\, to characterize the ligand-rece ptor interaction networks between adjacent tumor and immune/stroma cells based on ST data. LRnetST utilizes a directed acyclic graph model with a novel approach to handle the zero-inflated distributions of ST data. It a lso leverages existing ligand-receptor regulation databases as prior info rmation and employs a bootstrap aggregation strategy to achieve robust ne twork estimation. Application of LRnetST to ST data of high-grade serous ovarian tumor samples revealed both common and distinct ligand-receptor r egulations across different tumors. Some of these interactions were valid ated through both a MERFISH dataset and a CosMx SMI dataset of independen t ovarian tumor samples. These results cast light on biological processes relating to the communication between tumor and immune/stromal cells in ovarian tumors.\n\nAdmission:\nFree\n\nDetails URL:\nhttps://medicine.yal e.edu/event/ysph-biostatistics-seminar-3/\n DTEND;TZID=America/New_York:20260127T130000 DTSTAMP:20260517T040622Z DTSTART;TZID=America/New_York:20260127T120000 GEO:41.302961;-72.931638 LOCATION:106 A&B\, 47 College Street\, New Haven\, CT\, United States RECURRENCE-ID;TZID=America/New_York:20260127T120000 SEQUENCE:0 STATUS:Confirmed SUMMARY:YSPH Biostatistics Seminar: “Learning directed acyclic graphs for ligands and receptors based on spatially resolved transcriptomic data of ovarian cancer” UID:f0a07cbc-40b8-4d98-ab72-aac4972c18b7 END:VEVENT BEGIN:VEVENT DESCRIPTION:Note: BIS 526 students are required to attend in person. Other s are invited to attend in person but may also attend via zoom. Speaker J oseph W. Hogan\, Sc.D. Title "Bayesian clinical decision support for HIV care: development\, implementation\, evaluation." Abstract Key milestones in HIV care are represented in a cascade framework that prioritizes case identification\, linkage and retention in care\, compliance with antivir al medication\, and suppression of viral load. Many prediction models hav e been developed to target these endpoints. In this talk\, we describe a comprehensive approach to development\, implementation\, and formal evalu ation of a machine-learning-based approach to clinical decision support t hat is focused on retention in care. We highlight key early findings abou t effectiveness\, and describe methodologic innovations motivated by the goals of the project. This work is based at AMPATH\, a large HIV care pro gram in western Kenya.\n\nAdmission:\nFree\n\nDetails URL:\nhttps://medic ine.yale.edu/event/ysph-biostatistics-seminar-4/\n DTEND;TZID=America/New_York:20260203T130000 DTSTAMP:20260517T040622Z DTSTART;TZID=America/New_York:20260203T120000 GEO:41.302961;-72.931638 LOCATION:106 A&B\, 47 College Street\, New Haven\, CT\, United States RECURRENCE-ID;TZID=America/New_York:20260203T120000 SEQUENCE:0 STATUS:Confirmed SUMMARY:YSPH Biostatistics Seminar- Bayesian clinical decision support for HIV care: development\, implementation\, evaluation. UID:f0a07cbc-40b8-4d98-ab72-aac4972c18b7 END:VEVENT BEGIN:VEVENT DESCRIPTION:Note: BIS 526 students are required to attend in person. Other s are invited to attend in person but may also attend via zoom. Speaker R ui Feng\, Ph.D. Title: “Time-Varying Treatment Effects and Redesign of St epped-Wedge Cluster Randomized Trials” Abstract Stepped-wedge cluster ran domized trials introduce interventions sequentially across clusters and c ommonly assume constant treatment effects over time. In practice\, treatm ent effects often vary with time\, and violations of this homogeneity ass umption can lead to biased inference. In this talk\, I establish identifi ability conditions for time-varying\, piecewise treatment effects in two- arm and three-arm stepped-wedge trials and show how conventional designs yield biased treatment and time-effect estimates under misspecification. Simulations demonstrate inflated Type I error and bias in both main and i nteraction effects. I then illustrate the practical implications using th e DROP-BENZO trial\, showing that estimates under minimal homogeneity and limited time-adjustment assumptions are more plausible than those obtain ed under full homogeneity.\n\nAdmission:\nFree\n\nDetails URL:\nhttps://m edicine.yale.edu/event/ysph-biostatistics-seminar-5/\n DTEND;TZID=America/New_York:20260210T130000 DTSTAMP:20260517T040622Z DTSTART;TZID=America/New_York:20260210T120000 GEO:41.302961;-72.931638 LOCATION:106 A&B\, 47 College Street\, New Haven\, CT\, United States RECURRENCE-ID;TZID=America/New_York:20260210T120000 SEQUENCE:0 STATUS:Confirmed SUMMARY:YSPH Biostatistics Seminar- “Time-Varying Treatment Effects and Re design of Stepped-Wedge Cluster Randomized Trials” UID:f0a07cbc-40b8-4d98-ab72-aac4972c18b7 END:VEVENT BEGIN:VEVENT DESCRIPTION:Note: BIS 526 students are required to attend in person. Other s are invited to attend in person but may also attend via zoom. Speaker- Susan A. Murphy\, Ph\,D. Title- "Reinforcement Learning for Digital Healt h Interventions in the Dyadic Setting" Abstract We present our ongoing wo rk on the development of an online reinforcement learning (RL) algorithm for dyadic digital intervention settings in which the task for the RL alg orithm is to assist the target person with a difficult illness be adheren t to behavioral activities. To achieve this goal the RL algorithm will no t only deliver digital interventions to the target person but also delive r interventions to assist the care partner to manage caregiving burden an d help the two individuals improve their relationship. That is\, differen t RL components target different elements of the dyad. The RL algorithm i s a multi-agent RL algorithm in which the 3 agents make decisions on the 3 elements of the dyad. We incorporate domain knowledge in the form of ap proximal causal directed acyclic graphs to speed up online learning in th is sparse data setting. This work is motivated by our development of the ADAPTS-HCT multi-agent RL algorithm\, designed to improve medication adhe rence by young adults who have undergone a blood and bone marrow transpla nt. The RL algorithm will be deployed in summer 2026.\n\nAdmission:\nFree \n\nDetails URL:\nhttps://medicine.yale.edu/event/ysph-biostatistics-semi nar-6/\n DTEND;TZID=America/New_York:20260217T130000 DTSTAMP:20260517T040622Z DTSTART;TZID=America/New_York:20260217T120000 GEO:41.302961;-72.931638 LOCATION:106 A&B\, 47 College Street\, New Haven\, CT\, United States RECURRENCE-ID;TZID=America/New_York:20260217T120000 SEQUENCE:0 STATUS:Confirmed SUMMARY:YSPH Biostatistics Seminar-"Reinforcement Learning for Digital Hea lth Interventions in the Dyadic Setting" UID:f0a07cbc-40b8-4d98-ab72-aac4972c18b7 END:VEVENT BEGIN:VEVENT DESCRIPTION:Note: BIS 526 students are required to attend in person. Other s are invited to attend in person but may also attend via zoom. Speaker S ebastien Haneuse\, Ph.D. Title- Double sampling and semiparametric method s for informatively missing not at random data Abstract Missing data aris e in almost all applied settings and are ubiquitous in electronic health records (EHR). When data are missing not at random (MNAR) with respect to measured covariates\, sensitivity analyses are often considered. These s olutions\, however\, are unsatisfying in that they are not guaranteed to yield actionable conclusions. Motivated by an EHR-based study of long-ter m outcomes following bariatric surgery\, we consider the use of double sa mpling as a means to mitigate MNAR outcome data when the statistical goal s are estimation and inference regarding causal contrasts based on mean c ounterfactuals. We describe identification assumptions and derive efficie nt and robust estimators of the average causal treatment effect under a n onparametric model as well as under a model assuming the missing outcomes were\, in fact\, initially missing at random (MAR). We compare these in simulations to an approach that adaptively estimates based on evidence of violation of the MAR assumption. Finally\, we show how the methods can b e extended to: (i) estimation/inference regarding causal quantile treatme nt effects\; and (ii) hypothesis testing regarding MNAR.\n\nAdmission:\nF ree\n\nDetails URL:\nhttps://medicine.yale.edu/event/ysph-biostatistics-s eminar-8/\n DTEND;TZID=America/New_York:20260303T130000 DTSTAMP:20260517T040622Z DTSTART;TZID=America/New_York:20260303T120000 GEO:41.302961;-72.931638 LOCATION:106 A&B\, 47 College Street\, New Haven\, CT\, United States RECURRENCE-ID;TZID=America/New_York:20260303T120000 SEQUENCE:0 STATUS:Confirmed SUMMARY:YSPH Biostatistics Seminar- Double sampling and semiparametric met hods for informatively missing not at random data UID:f0a07cbc-40b8-4d98-ab72-aac4972c18b7 END:VEVENT BEGIN:VEVENT DESCRIPTION:Note: BIS 526 students are required to attend in person. Other s are invited to attend in person but may also attend via zoom. Speaker- Robert Krafty\, Ph.D. Title- Functional Data Analysis (FDA) under Informa tive Missingness with Application to the Analysis of Ecological Monetary Assessment (EMA) Abstract There has been an explosion in the use of mobil e device-based applications by clinicians and researchers to prompt indiv iduals to report on their mental and physical states in real time and the ir natural environments. Such observational data are inherently subject t o informative non-monotone missingness in which individuals do not reply to some prompts and where failing to reply is associated with the mental or physical states of interest. Although methods for the nonparametric an alysis of trajectories of longitudinal data have been well studied for de cades\, methods that are unbiased and efficient in the presence of inform ative non-monotone missingness are dearth. In this talk\, we introduce a joint model for functional outcomes with nonignorable missing values wher e the probability of missingness is a function of an interpretable differ ential operator of subject-specific functional effects. A Sobolev space-b ased EM algorithm is developed for penalized maximum likelihood-based est imation and inference. We discuss the consistency and efficiency of the e stimation procedure and illustrate its empirical properties in simulation studies. The practical implications of the method are illustrated throug h the analysis of negative affect in a study of young adults at risk for suicidal behavior.\n\nSpeaker:\nRobert Krafty\, Ph.D.\n\nAdmission:\nFree \n\nDetails URL:\nhttps://medicine.yale.edu/event/ysph-biostatistics-semi nar-11/\n DTEND;TZID=America/New_York:20260324T130000 DTSTAMP:20260517T040622Z DTSTART;TZID=America/New_York:20260324T120000 GEO:41.302961;-72.931638 LOCATION:106 A&B\, 47 College Street\, New Haven\, CT\, United States RECURRENCE-ID;TZID=America/New_York:20260324T120000 SEQUENCE:0 STATUS:Confirmed SUMMARY:YSPH Biostatistics Seminar- Functional Data Analysis (FDA) under I nformative Missingness with Application to the Analysis of Ecological Mon etary Assessment (EMA) UID:f0a07cbc-40b8-4d98-ab72-aac4972c18b7 END:VEVENT BEGIN:VEVENT DESCRIPTION:Note: BIS 526 students are required to attend in person. Other s are invited to attend in person but may also attend via zoom. Speaker- Qi Long\, Ph.D. Title- “Advancing Responsible Statistical and AI/ML Metho ds for Harnessing the Power of Electronic Health Records” Abstract Rich e lectronic health records (EHR) data offer remarkable opportunities in adv ancing precision medicine (Orcutt et al.\, 2025\, Nature Medicine )\, the y also present daunting analytical challenges. Multi-modal data in EHR th at are recorded at irregular time intervals with varying frequencies incl ude structured data such as labs and vitals\, codified data such as diagn osis and procedure codes\, and unstructured data such as clinical notes a nd pathology reports. They are typically incomplete and fraught with othe r errors and biases. What’s more\, data gaps and errors in EHRs are often unequally distributed across patient groups: People with less access to care\, often people with lower socioeconomic status\, tend to have more i ncomplete data in EHRs. Such data issues\, if not adequately addressed\, would lead to biased results and exacerbate health disparities (Getzen et al. 2023\, JBI ). In this talk\, I will share my research group’s recent works on developing responsible statistical and AI/ML methods including large language models (LLMs) and agentic AI for addressing these challeng es. Since LLMs are themselves plagued by various biases\, I will also dis cuss our ongoing research on developing rigorous statistical and ML appro aches for mitigating pitfalls and risks of LLMs (e.g.\, Xiao et al. 2025a JASA and 2025b\, ICML \; Li et al. 2025a AoS and 2025b JRSSB ).\n\nAdmis sion:\nFree\n\nDetails URL:\nhttps://medicine.yale.edu/event/ysph-biostat istics-seminar-12/\n DTEND;TZID=America/New_York:20260331T130000 DTSTAMP:20260517T040622Z DTSTART;TZID=America/New_York:20260331T120000 GEO:41.302961;-72.931638 LOCATION:106 A&B\, 47 College Street\, New Haven\, CT\, United States RECURRENCE-ID;TZID=America/New_York:20260331T120000 SEQUENCE:0 STATUS:Confirmed SUMMARY:YSPH Biostatistics Seminar- “Advancing Responsible Statistical and AI/ML Methods for Harnessing the Power of Electronic Health Records” UID:f0a07cbc-40b8-4d98-ab72-aac4972c18b7 END:VEVENT BEGIN:VEVENT DESCRIPTION:Note: BIS 526 students are required to attend in person. Other s are invited to attend in person but may also attend via zoom. Speaker- Jiayang Sun\, Ph.D. Title- "Learning from shifted data via a semiparametr ic selection bias modeling” Jiayang Sun1\, Zixiang Xu1\, Mary Meyer2\, Ji ng Qin3 1Department of Statistics\, George Mason University 2Colorado Sta te University\, 3NIH Abstract Data shifts occur in various forms\, such a s changes in covariates\, labels or prior distributions\, or domains from the first stage. This talk introduces recent advances in analyzing data with potential selection bias and explores how these relate to broader da ta-shift frameworks\, including co-moving shifts. We present the estimati on and testing procedures and demonstrate their effectiveness through the ory\, simulation studies\, and applications to large astronomical and/or heart-transplant datasets\, time permitting. Keywords: Data shifts\, sele ction bias\, isotonic inference\, smoothing splines\, heart transplant\, SDSS.\n\nSpeaker:\nJiayang Sun\, Ph.D.\n\nAdmission:\nFree\n\nDetails URL :\nhttps://medicine.yale.edu/event/ysph-biostatistics-seminar-13/\n DTEND;TZID=America/New_York:20260407T130000 DTSTAMP:20260517T040622Z DTSTART;TZID=America/New_York:20260407T120000 GEO:41.302961;-72.931638 LOCATION:106 A&B\, 47 College Street\, New Haven\, CT\, United States RECURRENCE-ID;TZID=America/New_York:20260407T120000 SEQUENCE:0 STATUS:Confirmed SUMMARY:YSPH Biostatistics Seminar- “Learning from shifted data via a semi parametric selection bias modeling” UID:f0a07cbc-40b8-4d98-ab72-aac4972c18b7 END:VEVENT BEGIN:VEVENT DESCRIPTION:Note: BIS 526 students are required to attend in person. Other s are invited to attend in person but may also attend via zoom. Speaker- Danyu Liu\, Ph.D. Title- "Evaluating the Effectiveness of COVID-19 Vaccin es Over Time” Abstract Approximately 800 million COVID-19 cases and 7 mil lion COVID-19 deaths have been reported to the World Health Organization thus far. Vaccination is a major tool to combat the COVID-19 pandemic\, b ut its effectiveness wanes over time and tends to be lower against new SA RS-CoV-2 variants. The knowledge about the waning effects of vaccination can guide boosting strategies. In a series of papers published in The New England Journal of Medicine and JAMA \, we reported several large cohort studies using COVID-19 case surveillance and vaccination data from the s tates of North Carolina and Nebraska. We developed a novel statistical fr amework to evaluate the time-varying effects of four generations of COVID -19 vaccines produced in the United States on infections with different S ARS-CoV-2 variants and on severe outcomes (hospitalization and death). Ou r findings have been cited by the World Health Organization and the U.S. Centers for Disease Control and Prevention and Food and Drug Administrati on and reported by The New York Times \, The Washington Post \, ABC News \, and NBC News . References Lin DY\, Gu Y\, Wheeler B\, Young H\, Hollow ay S\, Sunny SK\, Moore Z\, Zeng D (2022). Effectiveness of Covid-19 vacc ines over a 9-month period in North Carolina. New England Journal of Medi cine 386: 933-941. https://www.nejm.org/doi/full/10.1056/NEJMoa2117128 Li n DY\, Gu Y\, Xu Y\, Wheeler B\, Young H\, Sunny SK\, Moore Z\, Zeng D (2 022). Association of primary and booster vaccination and prior infection with SARS-CoV-2 infection and severe COVID-19 outcomes. JAMA 328: 1415-14 26. https://jamanetwork.com/journals/jama/fullarticle/2796893 Lin DY\, Gu Y\, Xu Y\, Zeng D\, Wheeler B\, Young H\, Sunny SK\, Moore Z (2022). Eff ects of vaccination and previous infection on omicron infections in child ren. New England Journal of Medicine 387: 1141-1143. https://www.nejm.org /doi/full/10.1056/NEJMc2209371 Lin DY\, Xu Y\, Gu Y\, Zeng D\, Wheeler B\ , Young H\, Sunny SK\, Moore Z (2023). Effectiveness of bivalent boosters against severe omicron infection. New England Journal of Medicine 388: 7 64-766. https://www.nejm.org/doi/full/10.1056/NEJMc2215471 Lin DY\, Du Y\ , Xu Y\, Paritala S\, Donahue M\, Maloney P (2024). Durability of XBB.1.5 vaccines against omicron subvariants. New England Journal of Medicine 39 0:2124-2127. https://www.nejm.org/doi/full/10.1056/NEJMc2402779 Du Y\, Pa ritala S\, Xu Y\, Maloney P\, Lin DY. Durability of 2024-2025 COVID-19 va ccines against JN. 1 subvariants (with commentary by Robert M Califf). JA MA Internal Medicine 185:1501-1504. https://jamanetwork.com/journals/jama internalmedicine/fullarticle/2840565\n\nAdmission:\nFree\n\nDetails URL:\ nhttps://medicine.yale.edu/event/ysph-biostatistics-seminar-14/\n DTEND;TZID=America/New_York:20260414T130000 DTSTAMP:20260517T040622Z DTSTART;TZID=America/New_York:20260414T120000 GEO:41.302961;-72.931638 LOCATION:106 A&B\, 47 College Street\, New Haven\, CT\, United States RECURRENCE-ID;TZID=America/New_York:20260414T120000 SEQUENCE:0 STATUS:Confirmed SUMMARY:YSPH Biostatistics Seminar- “Evaluating the Effectiveness of COVID -19 Vaccines Over Time” UID:f0a07cbc-40b8-4d98-ab72-aac4972c18b7 END:VEVENT BEGIN:VEVENT DESCRIPTION:Note: BIS 526 students are required to attend in person. Other s are invited to attend in person but may also attend via zoom. Speaker K un Chen\, Ph.D. Title- “Transfer Learning under Real-World Heterogeneity: Across-Site Contamination” Abstract Transfer learning and data fusion ai m to improve prediction for a target population by borrowing information from related sources. There has been an increasing need for such methods in multi-site EHR networks and health information exchange (HIE) settings . In this talk\, we discuss transfer learning under real-world heterogene ity\, which can be either a friend or a foe. As a friend\, heterogeneity is structured : sources and the target form clusters of similarity. In th is regime\, we present a three-stage\, cluster-aware transfer-learning pr ocedure that improves the bias–variance tradeoff and demonstrate it in fa cility-specific suicide risk modeling across 27 Connecticut hospitals. As a foe\, heterogeneity is unstructured : external data may be large but c ontaminated by arbitrary outliers. In this regime\, we study how to selec t/sample from the external data\, deriving non-asymptotic theory that mak es explicit how performance depends on target size\, sampling rate\, sign al strength\, outlier magnitude\, and error tail behavior. The take-home message is simple: more is not necessarily better\, and how much you can borrow depends on how much you have.\n\nAdmission:\nFree\n\nDetails URL:\ nhttps://medicine.yale.edu/event/ysph-biostatistics-seminar-15/\n DTEND;TZID=America/New_York:20260421T130000 DTSTAMP:20260517T040622Z DTSTART;TZID=America/New_York:20260421T120000 GEO:41.302961;-72.931638 LOCATION:106 A&B\, 47 College Street\, New Haven\, CT\, United States RECURRENCE-ID;TZID=America/New_York:20260421T120000 SEQUENCE:0 STATUS:Confirmed SUMMARY:YSPH Biostatistics Seminar- “Transfer Learning under Real-World He terogeneity: Across-Site Contamination” UID:f0a07cbc-40b8-4d98-ab72-aac4972c18b7 END:VEVENT BEGIN:VEVENT DESCRIPTION:Note: BIS 526 students are required to attend in person. Other s are invited to attend in person but may also attend via zoom. Speaker- Ying Guo\, Ph.D. Title- "A Statistical AI Method for Learning Directed Co nnectomes and Uncovering Subpopulation Differences in Brain Organization" Abstract In recent years\, connectome-based research has become a centra l focus in neuroscience\, offering essential insights into brain organiza tion and advancing predictive modeling of cognitive\, behavioral\, and me ntal health outcomes. While most existing approaches focus on undirected brain connectivity\, they overlook the directionality and causal influenc es between brain regions. To address this limitation\, we propose a stati stical AI method for learning directed brain connectomes from neuroimagin g data. Our approach integrates principled statistical modeling with deep learning to infer sparse\, interpretable directed connectivity graphs th at characterize latent causal interactions across the brain. At the same time\, the method learns low-dimensional graph embeddings optimized for d ownstream prediction tasks\, including demographic attributes and clinica l phenotypes. The proposed method uncovers whole-brain directed connectiv ity patterns and reveals novel subpopulation-specific connectomic differe nces\, highlighting its potential to advance both mechanistic understandi ng and predictive modeling in neuroscience\n\nSpeaker:\nYing Guo\, Ph.D.\ n\nAdmission:\nFree\n\nDetails URL:\nhttps://medicine.yale.edu/event/ysph -biostatistics-seminar-16/\n DTEND;TZID=America/New_York:20260428T130000 DTSTAMP:20260517T040622Z DTSTART;TZID=America/New_York:20260428T120000 GEO:41.302961;-72.931638 LOCATION:106 A&B\, 47 College Street\, New Haven\, CT\, United States RECURRENCE-ID;TZID=America/New_York:20260428T120000 SEQUENCE:0 STATUS:Confirmed SUMMARY:YSPH Biostatistics Seminar- "A Statistical AI Method for Learning Directed Connectomes and Uncovering Subpopulation Differences in Brain Or ganization" UID:f0a07cbc-40b8-4d98-ab72-aac4972c18b7 END:VEVENT BEGIN:VEVENT DESCRIPTION:Note: BIS 526 students are required to attend in person. Other s are invited to attend in person but may also attend via zoom. Speaker- Ying Guo\, Ph.D. Title- "A Statistical AI Method for Learning Directed Co nnectomes and Uncovering Subpopulation Differences in Brain Organization" Abstract In recent years\, connectome-based research has become a centra l focus in neuroscience\, offering essential insights into brain organiza tion and advancing predictive modeling of cognitive\, behavioral\, and me ntal health outcomes. While most existing approaches focus on undirected brain connectivity\, they overlook the directionality and causal influenc es between brain regions. To address this limitation\, we propose a stati stical AI method for learning directed brain connectomes from neuroimagin g data. Our approach integrates principled statistical modeling with deep learning to infer sparse\, interpretable directed connectivity graphs th at characterize latent causal interactions across the brain. At the same time\, the method learns low-dimensional graph embeddings optimized for d ownstream prediction tasks\, including demographic attributes and clinica l phenotypes. The proposed method uncovers whole-brain directed connectiv ity patterns and reveals novel subpopulation-specific connectomic differe nces\, highlighting its potential to advance both mechanistic understandi ng and predictive modeling in neuroscience\n\nSpeaker:\nYing Guo\, Ph.D.\ n\nAdmission:\nFree\n DTEND;TZID=America/New_York:20260113T130000 DTSTAMP:20260517T040622Z DTSTART;TZID=America/New_York:20260113T120000 EXDATE:20260224T120000 EXDATE:20260310T120000 EXDATE:20260317T120000 GEO:41.302961;-72.931638 LOCATION:106 A&B\, 47 College Street\, New Haven\, CT\, United States RRULE:FREQ=WEEKLY;UNTIL=20260429T035959Z;BYDAY=TU SEQUENCE:0 STATUS:Confirmed SUMMARY:YSPH Biostatistics Seminar- "A Statistical AI Method for Learning Directed Connectomes and Uncovering Subpopulation Differences in Brain Or ganization" UID:f0a07cbc-40b8-4d98-ab72-aac4972c18b7 END:VEVENT END:VCALENDAR