When you have mastered numbers, you will in fact no longer be reading numbers, any more than you read words when reading books. You will be reading meanings.--W.E.B. Du Bois
Another casualty of "embedded" reporting visible when traveling to not only statistical meetings but medical conferences, health IT meetings, and even policy and health economic discussions--I am observing "war" from just one perspective, not seeing the entire context. These sessions are run by and for experts in the field not among ordinary people, asking questions of all sides. My vantage point has value, but it is hardly a 360 degree perspective. Unless I also engage with different member stakeholders. This reminds me of CME colleagues, or professionals from Pharma, or even agencies. It is hard to make a significant difference if you only represent a single perspective.
The Joint statistics meeting was a "numbers gal" dream. Look at it this way I may not be a botanist, but I am definitely a flower. I follow the rules and accept the philosophy that generated them. I admit that statisticians like to use terms like "Idiot-proof" and "lay-person" like commas when discussing the data landscape but minus the Sturm und Drang, I get their concern.
Now, to be clear, I have studied (and continue to study) statistical principles and debates around ideology and philosophy. Big data has hit everyone by surprise--statisticians included. We have different accessibility issues, data readiness, and above all ideas about the best way to model data for the most accurate insights.
I am still applying the insights gained to a few current projects that will be improved by alternative models and robust analyses but I wanted to share a TEDx session from one of the speakers at JSM2017. Listen to an application of machine learning and how it has been applied to improving patient outcomes.
Better Medicine Through Machine Learning
Faster medical treatment saves lives. Machine Learning is already saving lives, by scouring a multitude of patients’ data and comparing them to one patient’s health data to detect symptoms 12 to 24 hours sooner than a doctor could. “In many pressing medical problems, the answers to knowing whom to treat, when to treat, and what to treat with, might already be in your data” says Suchi Saria. Learn how Saria's work on developing Targeted Real-time Early Warning Systems is leading the way to save lives.
The extended discussion of the network sampling approach has useful applications to clinical trial recruitment. Think of people as nodes and edges or links as relationships between people. In studies of hard to recruit patients where they may be dispersed or heard-to-reach, tracing the links within networks can be a logical method to add new numbers. This is an example of an adaptive sampling design because your sample will be selected based on the values of the nodes and links you observe during your survey. This has been my philosophy behind an extensive use of auxiliary variables.