The Problem
A mid-sized healthcare company wanted to be more targeted in their patient selection process so that they can maximize their impact through pharmacy consultation and reduce the hospitalization risk.
Our Approach
We built data models around medication and hospitalization history, demographic and consultation data. Predictive model was used to predict patient hospitalization risk with > 80% accuracy. We created a simulation model to simulate consultation impact. Model explanation framework was added to identify top reasons of high risk. Finally, we built visualizations via DOMO to track key metrics.
The Outcome
We improved patient selection accuracy by 100%+ and reduced false positive rate by 50%. The framework that’s been built is deployed into the operations on AWS and is actively being used every day.
“The Premier team is a delight to work with. World class skills across data prep, automation, BI, and ML/AI. Exceptional project management and client engagement. They have become integral members of our team. “
From Our Awesome Customer
The Problem
A financial service startup wanted to build machine learning model which would drive their whole business.
Our Approach
We became their strategic partner, understood the initial data and developed the machine learning model. We automated the process to pull the relevant and most recent data from APIs and rebuilt the engineering so that data can be pulled on daily basis and model predictions could be updated. We also re-engineered the website so that it can directly call this model for making predictions for user inputs.
The Outcome
The model developed had an accuracy of 80% and generated high-qualify predictions that helped the client increased their sales.
“The Premier Strategy team brought strong mathematical and machine learning skills to solve my particular project challenge. The team regularly communicated and progressed, allowing me to make progress quicker than expected!”
From Our Awesome Customer
The problem
A public biotech company wanted to maximize their growth chamber utilization rate. These growth chambers differ in sizes and client would like to experiment with many crops.
Our Apporach
We helped them collect all operational constraints by having discussions with key business stakeholders. We built an automated optimization tool using OR-Tools on AWS to optimize growth chamber testing scheduling.
The Outcome
Schedules using optimization tool build provide 34% higher growth chamber utilization rate. Each schedule takes a few hours to build, replacing weeks of manual work. Many scenario based questions can now be answered using our new tools, such as “could we reduce workload during weekends?”.