The Fred Hutch Biostatistics Program hosts seminars featuring presentations by Hutch and outside scientists to share their latest developments and recent research. Each seminar includes an hour-long presentation and discussion during which speakers showcase their work and findings.
This seminar will be held on Zoom due to the COVID-19 pandemic.
Biostatistics Seminar Series: “Linear biomarker combination under sensitivity or specificity constraint”
For disease detection, multiple biomarkers are often linearly combined for improved diagnostic performance. However, such combinations are typically obtained by employing convenient methods and/or performance metrics, resulting in sub-optimality except for restrictive circumstances. As a medical test usually needs to operate at a certain high sensitivity or specificity level, in this talk I will discuss some of our on-going developments to directly target the clinical utility by empirically optimizing the performance metric of specificity at a controlled sensitivity level or vice versa. Limiting distribution of the combination coefficients is established, leading to novel estimation and inference procedures to address over-estimation in the apparent performance estimation. An algorithm is devised for efficient and high-quality computation. Simulations demonstrate good statistical and computational performance. Illustration with a clinical study is provided.