University of California, San Francisco
Postdoctoral Researcher
University of California, San Francisco, San Francisco, California, United States, 94199
Overview
Postdoctoral Scholar in AI/ML and Causal Inference for Sepsis Decision Support The Department of Anesthesia and Perioperative Care at UCSF and the EpochAI lab are seeking a highly motivated Postdoctoral Scholar with strong expertise in machine learning (ML) and causal inference to join an NIH R01-funded project focused on improving early treatment decisions for patients with community-onset lung sepsis (COLS). This is a unique opportunity to contribute to the development of a real-time, EHR-integrated clinical decision support system (CDSS) designed to estimate individual treatment effects (ITE) of sepsis therapies (antibiotics, fluids, vasopressors) and to personal ize care based on data-driven insights. You will work within a multidisciplinary team of clinicians, data scientists, implementation scientists, and health systems researchers across UCSF, UPenn, and UPMC. Responsibilities
Develop and validate machine learning models to estimate counterfactual outcomes and individual treatment effects (CATE/ITE) using EHR data Apply advanced modeling techniques including Highly Adaptive Lasso, Generative Adversarial Networks (GANs), and other interpretable ML methods Causal inference frameworks (e.g., T-Learner, DR-Learner, G-computation) Collaborate with clinical teams to translate model output into actionable CDSS recommendations Contribute to the design and testing of an EHR-embedded user interface for the CDSS Lead and contribute to scientific publications, presentations, and progress reports Participate in regular project meetings and collaborative data science discussions Support simulation and net benefit analyses to evaluate the clinical utility of the CDSS Required Qualifications
PhD (or equivalent) in computer science, statistics, epidemiology, biomedical informatics, or related field Demonstrated experience in machine learning, causal inference, and/or clinical prediction modeling Proficiency in Python and/or R, and experience working with large-scale clinical data Familiarity with counterfactual reasoning and model evaluation metrics (e.g., calibration, discrimination, bias) Strong communication and scientific writing skills Ability to work independently and collaboratively in a fast-paced, interdisciplinary environment Preferred Qualifications
Experience working with electronic health records (EHR) or claims data Familiarity with implementation science or human-centered design Previous work related to sepsis, critical care, or clinical decision support systems Knowledge of Bayesian models, bootstrap estimation, or semi-supervised learning Duration: 1–2 years with potential for extension based on performance and funding Start date: Flexible, with preference for Fall/Winter 2025 Salary: Commensurate with UCSF postdoctoral scale and experience Application Instructions
To apply, please send the following to [PI Email]: CV Cover letter describing your interest in the position and relevant experience Contact information for 2–3 references
#J-18808-Ljbffr
Postdoctoral Scholar in AI/ML and Causal Inference for Sepsis Decision Support The Department of Anesthesia and Perioperative Care at UCSF and the EpochAI lab are seeking a highly motivated Postdoctoral Scholar with strong expertise in machine learning (ML) and causal inference to join an NIH R01-funded project focused on improving early treatment decisions for patients with community-onset lung sepsis (COLS). This is a unique opportunity to contribute to the development of a real-time, EHR-integrated clinical decision support system (CDSS) designed to estimate individual treatment effects (ITE) of sepsis therapies (antibiotics, fluids, vasopressors) and to personal ize care based on data-driven insights. You will work within a multidisciplinary team of clinicians, data scientists, implementation scientists, and health systems researchers across UCSF, UPenn, and UPMC. Responsibilities
Develop and validate machine learning models to estimate counterfactual outcomes and individual treatment effects (CATE/ITE) using EHR data Apply advanced modeling techniques including Highly Adaptive Lasso, Generative Adversarial Networks (GANs), and other interpretable ML methods Causal inference frameworks (e.g., T-Learner, DR-Learner, G-computation) Collaborate with clinical teams to translate model output into actionable CDSS recommendations Contribute to the design and testing of an EHR-embedded user interface for the CDSS Lead and contribute to scientific publications, presentations, and progress reports Participate in regular project meetings and collaborative data science discussions Support simulation and net benefit analyses to evaluate the clinical utility of the CDSS Required Qualifications
PhD (or equivalent) in computer science, statistics, epidemiology, biomedical informatics, or related field Demonstrated experience in machine learning, causal inference, and/or clinical prediction modeling Proficiency in Python and/or R, and experience working with large-scale clinical data Familiarity with counterfactual reasoning and model evaluation metrics (e.g., calibration, discrimination, bias) Strong communication and scientific writing skills Ability to work independently and collaboratively in a fast-paced, interdisciplinary environment Preferred Qualifications
Experience working with electronic health records (EHR) or claims data Familiarity with implementation science or human-centered design Previous work related to sepsis, critical care, or clinical decision support systems Knowledge of Bayesian models, bootstrap estimation, or semi-supervised learning Duration: 1–2 years with potential for extension based on performance and funding Start date: Flexible, with preference for Fall/Winter 2025 Salary: Commensurate with UCSF postdoctoral scale and experience Application Instructions
To apply, please send the following to [PI Email]: CV Cover letter describing your interest in the position and relevant experience Contact information for 2–3 references
#J-18808-Ljbffr