Hacknosis: Data Analysis Winner!

A George

Project: Patient Graph

Team: ravi4555 (Ravi Bajracharya, Jans Aasman)

Description

Leading industry experts, including Gartner, concur that Knowledge Graphs are pivotal for contemporary data and analytics. They appear to offer the most effective solution for discerning relationships among diverse data sets. In the healthcare context, the Knowledge Graph approach addresses the complexity issue by streamlining 18,000 tables and 200,000 columns down to 350 classes and 1,000 attributes. This optimization offers data scientists a consistent 'data shape' applicable across various data science and reporting tasks. By viewing a patient both as a sequence of events and as a graph, data scientists can better pinpoint high-cost, high-need patients, enabling healthcare organizations to provide more timely and targeted care. By leveraging the Knowledge Graph methodology, an average-sized hospital system stands to save between $10 to $20 million annually. 

Patient Graph solution is designed to fast track a healthcare organization’s adoption of the Knowledge Graph approach and quickly begin to improve outcomes and lower costs of preventable diseases. Key features:

• Model healthcare data as entity-event knowledge graphs and solve the data complexity problem. 
• Solve data incompleteness and inaccuracy problem using OT AI to extract medical terms from unstructured data (medical notes, clinical trials, PubMed articles, VAERS events) and link them to UMLS taxonomy.
• With knowledge graph, generate analytics such as predicting readmissions, or finding patients with untreated diabetes. 
• Provides a graphical representation of the analytics on a dashboard

Judges thought this was a unique way to represent and solve this challenge. The insights from this team's first hand experience lead to a creative take on solving a complex problem and the judges loved the inclusion and use of OpenText's Capture, Metadata and Storage API's in this project.