Date: Tuesday, December 3, 2024 at 12:00 pm - 1:00 pm ET
The COPSS-NISS Leadership Webinar: Leadership at the Intersection of Causal Inference and Interdisciplinarity focused on leadership at the intersection of causal inference and interdisciplinarity, featuring two outstanding leaders, Dr. Mark Van der Laan and Dr. Elizabeth (Liz) Stewart, who have made significant contributions to the field of causal inference. The discussion covered Mark's research journey, his ongoing work on a book about the depth of Lassau and its role in targeted learning, and Elizabeth's work in design and analysis methods for estimating population causal effects. The conversation ended with a discussion on the importance of understanding and applying causal inference in their work, particularly in relation to AI and machine learning.
Causal Inference and Interdisciplinary Leadership
Shu Yang, an associate professor of statistics at NC State and a member of the 2024 class of the COPSS Leadership Academy, hosted a webinar on 'Leadership at the intersection of causal inference and interdisciplinarity'. The webinar featured two outstanding leaders, Dr. Mark Van der Laan and Dr. Liz Stewart, who have made significant contributions to the field of causal inference. The discussion focused on methodological advances, challenges in real-world applications, and the leadership required to bridge diverse fields. Shu introduced Mark Van der Laan, a renowned professor of biostatistics at UC Berkeley, who shared his journey of research, primarily focused on learning optimally and robustly from complex real-world data. The webinar aimed to provide insights into how participants' experiences could inspire their own work.
Mark's Research Journey and Contributions
Mark discussed his research journey, starting as a PhD student under Richard Gill, where he learned about counting processes and multiplicative intensity models. He then worked on generalizing the Kepler-Meyer for rights and affiliate time data to learn the Milford survival function. Mark also learned about the theory of establishing the asymptotic optimality of these estimators and the functional delta method. He collaborated with Jamie Robbins on complex observational studies and wrote a book on unified methods for centered longitudinal data and causality. Mark also delved into machine learning, studying cross-validation and proving an oracle inequality. He introduced the concept of the super learner and worked on multiple testing in genomics. Mark also developed targeted maximum likelihood estimation and adaptive designs, and he developed highly depth loss as a general, powerful algorithm. Mark discussed his ongoing work on a book about the depth of Lassau and its role in targeted learning, which is expected to be completed next year. He also shared his research interests, which include applications funded by NIH grants, genomics, environmental studies, and diabetes. Additionally, he mentioned his involvement in a center for target machine learning at Berkeley and his interest in FDA-related applications.
Causal Inference Methods and Challenges
Our Moderator Shu then introduced Elizabeth, who shared her journey in using data and evidence to make the world a better place. She highlighted her interdisciplinary work in causal inference and statistics, her time at Mathematica policy research, and her current role as chair of the Department of Biostatistics at Johns Hopkins University. Elizabeth discussed her work in design and analysis methods for estimating population causal effects, focusing on internal and external validity. She highlighted her experience at Mathematica and the challenges of conducting randomized trials. Elizabeth also emphasized the importance of interdisciplinary collaboration, particularly in public health, and the need for strong causal inference methods. She suggested that future developments in causal inference should focus on design approaches, diagnostics, and sensitivity analyses. Mark and Shu discussed the importance of understanding the assumptions and factors involved in causal inference, particularly in longitudinal data. They also touched on the need for real-world evidence and the challenges of generating it.
Precision Targeting in Academic Publishing
Mark discussed the importance of precise targeting and the need for reliable methods in academic publishing. He emphasized the significance of setting benchmarks and guidelines for a stepwise process that generates real evidence. Shu asked for specific examples of impactful problems to work on, to which Elizabeth responded, highlighting the importance of collaborations where all parties understand the need for a PhD-level statistician. She also stressed the importance of public health impact and the need for collaborators to empower statisticians to lead projects. Elizabeth and Mark agreed on the importance of collaborations where all parties are excited about finding the right approach for answering real questions of interest. Shu then asked about challenges faced in these collaborative projects, to which Elizabeth responded, mentioning collaborators who don't see the value in statisticians and those who want to ask questions that the data is not feasible to answer.
Balancing Research and Practicality
Elizabeth discussed the challenges of helping people understand the limitations of their research questions and the need for feasible projects. She also shared her experiences with publishing in top scientific journals and dealing with editors' concerns about new methods. Mark agreed with Elizabeth's points and emphasized the importance of using methods supported by the data, even if it means starting with simpler approaches. He also stressed the need for a balance between theoretical development and practical applicability. Shu then asked for further discussion on how to strike this balance.
Valuing Diverse Contributions in Biostatistics
Elizabeth and Mark discussed the importance of valuing different contributions in the field of biostatistics, emphasizing the need for a diverse range of skills and interests. They highlighted the significance of mentorship and leadership in fostering innovation and collaboration within teams, particularly in guiding individuals towards career paths that align with their strengths. They also stressed the importance of creating a supportive and open environment where team members feel comfortable asking questions and learning from each other. Both agreed on the value of collaborations between researchers at similar career stages and the need for honesty and transparency in their work.
Causal Inference in AI and ML
In the meeting, Mark and Elizabeth discussed the importance of understanding and applying causal inference in their work, particularly in relation to AI and machine learning. They emphasized the need for careful consideration of assumptions and the potential for AI to mislead without proper understanding of causal effects. They also highlighted the importance of clear communication with collaborators who may not have a background in causal inference, suggesting a step-by-step approach to explaining causal relationships. The conversation ended with a discussion on the potential for AI and machine learning to enhance causal inference, with Mark sharing examples of successful applications.
Thanks and Recognition
We extend our deepest gratitude to Dr. Mark Van der Laan and Dr. Elizabeth (Liz) Stewart for being our speakers in the COPSS-NISS Leadership Webinar: Leadership at the Intersection of Causal Inference and Interdisciplinarity. Their engaging discussion on methodological advances, interdisciplinary collaboration, and leadership provided invaluable insights for researchers navigating this complex field.
Special thanks to Dr. Shu Yang, who expertly moderated the session, guiding thoughtful conversations about the challenges and opportunities in applying causal inference across diverse domains, including AI and machine learning. Her leadership and dedication to fostering dialogue have made this webinar a true success.
About the COPSS-NISS Leadership Webinar Series
COPSS (Committee of the Presidents of Statistical Societies) and NISS have come together to organize and host a new webinar series focusing on leadership in statistics and data science. Plan to attend these webinars every month during the academic year! Visit the COPSS-NISS Leadership Series Page for previous webinars.
The COPSS-NISS Leadership Webinar Series is co-organized by the Committee of the Presidents of Statistical Societies (COPSS) Emerging Leaders in Statistics and the National Institute of Statistical Sciences (NISS). The purpose of the webinar series is to promote leadership skills for members of the statistical societies at any stage in their careers. The series features conversations with leaders throughout the discipline, including leaders from major academic and government institutions, and companies. Invited speakers share their leadership stories and answer questions about their experiences. Each webinar is moderated by a member of the COPSS Emerging Leaders in Statistics program.
Access the Full COPSS-NISS Leadership Webinar Series YouTube Playlist | COPSS-NISS Leadership Webinar Series: https://www.youtube.com/playlist?list=PLoRtupvDJTjvFukMcO6NfDr0GvxDsIj81
Tuesday, December 3, 2024 by Megan Glenn