Proprietary Model Uses De-identified Clinical Data to Predict Which Patients Will Develop Specific Disease States
“Luckily the NLP Logix team has a lot of experience merging clinical and administrative data and putting it into an actionable format, which helped us in this competition.” – Matt Berseth, Chief Information Officer, NLP Logix
JACKSONVILLE, FLORIDA – OCTOBER 08, 2012
Practice Fusion, the nation’s largest doctor-patient community, has named Matthew Berseth, Chief Information Officer for NLP Logix, a Jacksonville-based healthcare information technology development and integration company, a winner in its Prediction Problem Challenge.
This challenge was hosted and scored on a technology platform provided by Kaggle, which uses its Prospect platform to create international competitions to solve problems around data exploration and analysis, visualization and problem identification. Berseth’s winning submission used both de-identified clinical data contained in an electronic health record as well as data derived from insurance claims to predict who has and who would acquire type 2 diabetes.
“Healthcare has a lot of rich data sources, but they tend to be in the disparate locations where care is provided. Taking those sources, merging them and helping to streamline the transitions of care is something we do well.” said Berseth. “Luckily the NLP Logix team has a lot of experience merging clinical and administrative data and putting it into an actionable format, which helped us in this competition.”
The competition attracted 170 participants from across the world who submitted 2,200 models. Along with Berseth, the other two prize winners were data scientists from Sevilla, Spain and Mumbai, India.