Innovation and solutions in evidence science

A "systems-thinking" perspective is crucial for the analytic design connecting the data-generating process with the inference-generating process.  Research with observational data (without formal experimental design and randomization) is conceived as transmission of information through a channel of communication (see the epistemologic model below).   By identifying the numerous sources of information loss or distortion (noise) in the health research process and applying the best solution set we can create a more efficient, reliable and productive evidence generation system.  This has very pragmatic implications for optimizing cost efficiency and value generation for any evidence generation endeavor; and ultimately, for improving outcomes for patients.

Current foci:

GoodScience is advising on modern applied methods for observational data analysis, using subject-matter based  structural causal models to guide model building

Longitudinal analysis methods for cohort studies

Design and evaluation of rigorous inference using external comparators

Simulation and prototyping of evidence generation to optimize evidence generation strategies  

• Thought leadership communicated in social media blog posts.  

Examples include:

Value-based Healthcare: analytics may be leaving value on the table

Why 'Why' Matters: "The Book of Why"

In Machine Learning Predictions for Health Care the Confusion Matrix is a Matrix of Confusion

Navigating Statistical Modeling and Machine Learning

Observational Data Analysis (ODA) material

Variation in Practice material

Value of Information Methods for Evidence Generation Strategies

Other miscellaneous work examples

Innovation and Strategy