Advancing the screening, diagnostics, and treatment of diseases frequently requires addressing a causal question in medicine and healthcare. For example, what effect does the introduction of artificial intelligence tools have on clinical workflows and patient outcomes? How do changes in insurance coverage affect access to care? The ability to provide credible evidence to address these questions has important implications for understanding and improving healthcare practices. However, addressing such questions in a rigorous way requires a critical understanding of causal inference methods, including when causal inference may not be feasible. This course provides an introduction to causal inference, providing both statistical foundations and practical applications. We will introduce methods across both experimental and observational settings, including the design and analysis of randomized controlled trials, quasi-experimental methods such as instrumental variables, and propensity score-based matching and weighting methods. Additionally, we will ground these methods in concrete examples in healthcare from the scientific literature and will provide interactive hands-on coding practice for analyzing real-world data. Class meetings will be structured in two parts, with the first part being traditional lecture-style and the second part being an in-person workshop where students will work through coding notebooks to apply the newly learned concepts with the instructors' support. Recommended prerequisites are an introductory statistics course at an undergraduate level and basic knowledge of R or Python.
1 units · Medical Satisfactory/No Credit
Advancing the screening, diagnostics, and treatment of diseases frequently requires addressing a causal question in medicine and healthcare. For example, what effect does the introduction of artificial intelligence tools have on clinical workflows and patient outcomes? How do changes in insurance coverage affect access to care? The ability to provide credible evidence to address these questions has important implications for understanding and improving healthcare practices. However, addressing such questions in a rigorous way requires a critical understanding of causal inference methods, including when causal inference may not be feasible. This course provides an introduction to causal inference, providing both statistical foundations and practical applications. We will introduce methods across both experimental and observational settings, including the design and analysis of randomized controlled trials, quasi-experimental methods such as instrumental variables, and propensity score-based matching and weighting methods. Additionally, we will ground these methods in concrete examples in healthcare from the scientific literature and will provide interactive hands-on coding practice for analyzing real-world data. Class meetings will be structured in two parts, with the first part being traditional lecture-style and the second part being an in-person workshop where students will work through coding notebooks to apply the newly learned concepts with the instructors' support. Recommended prerequisites are an introductory statistics course at an undergraduate level and basic knowledge of R or Python.
Offered in Autumn 2026 at Stanford University.