I’ve provided data science consulting services for BUILD and Bernoulli, among others.
BUILD
Delta Analytics
I love helping set up a data strategy, starting with the big question an organization is trying to answer.
Working with BUILD, I helped pinpoint their ‘North Star metric’, putting a spotlight on how their program helps students graduate high school and stay in college, not just the number of programs they’re running.
By digging into the data and keeping the key questions in mind, I was glad to highlight the impact BUILD was making and provide an automated report that connected directly with their data sources.
See more about this project in the Delta blog.
Bernoulli One
Wharton Digital Health Club
Early detection of critical events saves lives. Yet throwing too many alarms leads to alarm fatigue for doctors and nurses, which can result desensitization to alarms and missed alarms.
So, given hundreds of hours worth of continuous patient data streams for key indicators, how can we better understand the interaction of health indicators to detect critical events more accurately?
Here I applied data mining techiniques (time-series, PCA, decision trees, auto-correlations, cross-correlations) to identify which indicators were useful predictors of the factory standard health index, and which were triggers for the majority of alerts.
The result was very promising - I was able to reliably detect severe health events 12 seconds (two time units) ahead of the industry standard pulmonary index (IPI) maintaining the same level of accuracy.
Our team them applied this knowledge to define the trade-off of false alarms and true health alerts with this custom indicator vs other standards, and we earned a highly positive review Bernoulli executive team.
See more about Bernoulli at BernoulliHealth.com