|
Personalized medicine is increasingly recognized as an important
aspect of modern drug development in which treatments are
tailored to individual patients.
An important aspect of personalized medicine is to identify
subgroups of patients who respond more strongly to
treatment than others – either in terms of efficacy, or safety.
This step ought to be a critical stage
in the analysis of any Phase II study so that any relevant
patient subpopulations are identified accurately and systematically.
Moreover, if no such subpopulations exist, proper data mining
will convincingly establish the fact, enabling better informed
and more swift decisions.
Our standard approach to subgroup identification draws on
methodology from the fields of machine learning and causal
inference and has been successfully applied to Phase II and
Phase III data to identify both efficacy responders and subgroups
at increased risk of serious adverse events.
|
