Insight Extraction From E-Health Bookings by Means of Hypergraph and Machine Learning

Area of Science:

• Health Informatics
• Data Science
• Medical Data Management

Background:

• Public healthcare systems manage patient access through centralized booking.
• E-health data requires efficient organization and retrieval for improved services.
• Existing data structures may obscure valuable patient insights.

Purpose of the Study:

• To present a Knowledge Graph (KG) method for structuring e-health data.
• To support e-health services by extracting medical knowledge and insights.
• To explore the application of machine learning on KG-embedded health data.

Main Methods:

• Utilized raw health booking data from Italy's public healthcare system.
• Developed a Knowledge Graph (KG) approach to organize and analyze the data.
• Applied graph embedding techniques to map entity attributes into a vector space.
• Implemented unsupervised and supervised Machine Learning (ML) on embedded vectors.

Main Results:

• Knowledge Graphs (KGs) can effectively assess patient medical booking patterns.
• Unsupervised ML identified hidden entity groups not apparent in raw data.
• Supervised ML showed promising, though not high, performance in predicting annual medical visits.

Conclusions:

• Knowledge Graphs (KGs) offer a viable method for enhancing e-health services.
• KG-based analysis can reveal hidden patterns and aid in predicting patient behavior.
• Further advancements in graph databases and embedding algorithms are needed for optimal application.