BEGIN:VCALENDAR
VERSION:2.0
PRODID:-//Institute of Statistical Research and Training - ECPv6.15.9//NONSGML v1.0//EN
CALSCALE:GREGORIAN
METHOD:PUBLISH
X-WR-CALNAME:Institute of Statistical Research and Training
X-ORIGINAL-URL:https://isrt.ac.bd
X-WR-CALDESC:Events for Institute of Statistical Research and Training
REFRESH-INTERVAL;VALUE=DURATION:PT1H
X-Robots-Tag:noindex
X-PUBLISHED-TTL:PT1H
BEGIN:VTIMEZONE
TZID:UTC
BEGIN:STANDARD
TZOFFSETFROM:+0000
TZOFFSETTO:+0000
TZNAME:UTC
DTSTART:20240101T000000
END:STANDARD
END:VTIMEZONE
BEGIN:VEVENT
DTSTART;TZID=UTC:20250204T113000
DTEND;TZID=UTC:20250204T130000
DTSTAMP:20251026T105748
CREATED:20250201T174134Z
LAST-MODIFIED:20250201T174134Z
UID:7409-1738668600-1738674000@isrt.ac.bd
SUMMARY:Applied Statistics and Data Science Seminar on Tuesday February 4\, 2025
DESCRIPTION:Title: Imputation-Based Q-Learning for Optimizing Dynamic Treatment Regimes with Right-Censored Survival Outcome \nVenue\, time and date: ISRT\, 11:30 pm\, February 4\, 2025 \nSpeaker: Abdus S. Wahed\, PhD\, Professor of Biostatistics\, Department of Biostatistics and Computational Biology\, Deltin 7 Aviator গেম টাকা ইনকাম of Rochester\, USA \nAbstract: \nQ-learning has been one of the most commonly used methods for optimizing dynamic treatment regimes (DTRs) in multistage decision-making. Right-censored survival outcome poses a significant challenge to Q-Learning due to its reliance on parametric models for counterfactual estimation which are subject to misspecification and sensitive to missing covariates. In this paper\, we propose an imputation-based Q-learning (IQ-learning) where flexible nonparametric or semiparametric models are employed to estimate optimal treatment rules for each stage and then weighted hot-deck multiple imputation (MI) and direct-draw MI are used to predict optimal potential survival times. Missing data are handled using inverse probability weighting and MI\, and the nonrandom treatment assignment among the observed is accounted for using a propensity-score approach. We investigate the performance of IQ-learning via extensive simulations and show that it is more robust to model misspecification than existing Q-Learning methods\, imputes only plausible potential survival times contrary to parametric models and provides more flexibility in terms of baseline hazard shape. Using IQ-learning\, we developed an optimal DTR for leukemia treatment based on a randomized trial with observational follow-up that motivated this study. \nPaper link: \nImputation-Based Q-Learning for Optimizing Dynamic Treatment Regimes with Right-Censored Survival Outcome | Biometrics | Oxford Academic 
URL:https://isrt.ac.bd/event/applied-statistics-and-data-science-seminar-on-tuesday-february-4-2025/
CATEGORIES:seminar
END:VEVENT
BEGIN:VEVENT
DTSTART;TZID=UTC:20250224T140000
DTEND;TZID=UTC:20250224T150000
DTSTAMP:20251026T105748
CREATED:20250222T040452Z
LAST-MODIFIED:20250222T040711Z
UID:7476-1740405600-1740409200@isrt.ac.bd
SUMMARY:Applied Statistics and Data Science Seminar on Monday February 24\, 2025
DESCRIPTION:Venue\, time and date: ISRT\, 2:00 pm\, February 24\, 2025 \nSpeaker: Humayera Islam\, PhD\, Postdoctoral Scholar in Precision Health at the Deltin 7 Aviator গেম টাকা ইনকাম of Chicago \nTalk 1 \nTitle: From Statistical Models to LLMs: The Evolution of Feature Representation in Predictive Modeling \nAbstract: \nWith the digitization of healthcare and public health systems\, data collection has expanded far beyond traditional numerical and categorical formats to include complex modalities such as natural language (e.g.\, clinical notes)\, medical images (e.g.\, radiology scans)\, genetic data (e.g.\, omics)\, and temporally extensive time-series data (e.g.\, electronic health records). This expansion was driven by advancements in data storage capacity\, enabling the collection of massive\, high-dimensional datasets. As the size and complexity of data grew\, so did the need for more sophisticated feature representation techniques to effectively capture the underlying patterns for predictive tasks to enhance clinical decision making. This seminar traces the evolution of feature representation from traditional statistical models\, which relied on manual feature engineering\, to machine learning models that automated feature extraction\, to deep learning architectures that learned hierarchical and temporal features\, and finally to Large Language Models (LLMs) that leveraged self-attention mechanisms for contextual sequence modeling. The aim is to spark curiosity and inspire students to explore how to effectively handle these diverse data modalities and harness the power of advanced models for innovative research projects. \nTalk 2 \nTitle: Pathways to Growth: Preparing for Data Science and Informatics Graduate Programs in the US \nAbstract: \nThis talk offers a comprehensive roadmap for students aspiring to pursue graduate programs in data science and informatics in the US. It will cover the key skill sets essential for enhancing data science expertise\, including domain knowledge\, emerging methodologies\, technical proficiency\, and leadership in professional development. Additionally\, students will be provided with valuable resources such as open-source datasets and learning platforms to strengthen these skills during their time at ISRT and effectively prepare for graduate studies.
URL:https://isrt.ac.bd/event/applied-statistics-and-data-science-seminar-on-monday-february-24-2025/
CATEGORIES:seminar
END:VEVENT
END:VCALENDAR