Microsoft, UWash Use NLP to Detect ICU Pneumonia
Researchers at the University of Washington and Microsoft have developed technology to speed up the diagnosis of pneumonia in ICU patients.
Researchers at the University of Washington and Microsoft have developed technology that uses natural language processing and machine learning to speed up the diagnosis of pneumonia in ICU patients.
Bioinformatics professor Meliha Yetisgen and her colleagues at the university teamed up with Microsoft researcher Lucy Vanderwende on the project, called deCIPHER, using the Microsoft Research Statistical Parsing and Linguistic Analysis Toolkit (Splat).
Yetisgen says the researchers studied the diagnosis of pneumonia in approximately 100 patients being treated in the ICU at Seattles Harborview Medical Center. By using the electronic medical records of these patients, whose pneumonia diagnoses had already been established by clinical consensus, they employed NLP tools from Microsoft to identify the critical clinical information.
They then ran that data through a machine-learning framework to see if the software could be trained to correctly diagnose the pneumonia cases based solely on an automated review of the digital medical records.
The results were so promising the software achieved a correct diagnosis with correct time-of-onset for positive cases in 84 percent of the patients that the team's clinical collaborators are considering the addition of their pneumonia-detection models to the dashboard they use to monitor ICU patients, Yetisgen says.
Bioinformatics professor Meliha Yetisgen and her colleagues at the university teamed up with Microsoft researcher Lucy Vanderwende on the project, called deCIPHER, using the Microsoft Research Statistical Parsing and Linguistic Analysis Toolkit (Splat).
Yetisgen says the researchers studied the diagnosis of pneumonia in approximately 100 patients being treated in the ICU at Seattles Harborview Medical Center. By using the electronic medical records of these patients, whose pneumonia diagnoses had already been established by clinical consensus, they employed NLP tools from Microsoft to identify the critical clinical information.
They then ran that data through a machine-learning framework to see if the software could be trained to correctly diagnose the pneumonia cases based solely on an automated review of the digital medical records.
The results were so promising the software achieved a correct diagnosis with correct time-of-onset for positive cases in 84 percent of the patients that the team's clinical collaborators are considering the addition of their pneumonia-detection models to the dashboard they use to monitor ICU patients, Yetisgen says.