VisAGE: Integrating external knowledge into electronic medical record visualization.

Area of Science:

• Medical Informatics
• Data Visualization
• Computational Biology

Background:

• Electronic Medical Records (EMRs) are crucial for patient care but often incomplete or fragmented.
• Visualizing EMRs in a low-dimensional space can aid in identifying patient similarities and disease subtypes.
• Challenges exist in accurately representing patients with missing data in visualizations.

Purpose of the Study:

• To present VisAGE, a novel method for visualizing electronic medical records (EMRs) in a low-dimensional space.
• To address the issue of incomplete or fragmented EMR data in patient visualization.
• To enhance the identification of patient subtypes and reduce misdiagnosis.

Main Methods:

• VisAGE integrates multiple external data sources to enrich fragmented EMR databases.
• The method visualizes patient data in a low-dimensional space for intuitive analysis.
• Evaluation was performed on a dataset of Parkinson's disease patients.

Main Results:

• VisAGE effectively clusters patients, even with incomplete EMR data.
• The visualization method allows for better discovery of patient subtypes.
• Qualitative and quantitative analyses demonstrate improved patient clustering compared to existing methods.

Conclusions:

• VisAGE enhances the visualization of electronic medical records by integrating external data.
• Improved patient clustering facilitates better identification of disease subtypes.
• The VisAGE method has the potential to improve patient care and reduce diagnostic errors.