Clinical diagnostic devices
Development of diagnostic devices benefits from methods that support product and process design and optimization. Besides, there are a few capabilities specific to diagnostic devices:
- Design and analyze validation studies for regulatory submission
- Predictive modeling to inform critical system cutoffs
Product/process design
- “Design of Experiments” methods to identify influential variables and converge rapidly on a robust, optimum design
- Variance components analysis to understand what factors contribute to variability, and how much
Exploratory predictive modeling
- I’m adept with the modeling methods described by Frank Harrell, Jr.
- I’m familiar with a wide range of machine learning methods. I can advise on which is likely to be most effective for your particular data set, the particular need you’re addressing, and your audience.
- I’m familiar with clustering and population segmentation methods. I particularly recommend methods that use statistical modeling approaches, such as estimating density with splines or fitting a finite mixture model, because these tend to be more stable and use data more efficiently than purely algorithmic methods.