An ensemble approach for healthcare application and diagnosis using natural language processing

Area of Science:

• Health Informatics
• Artificial Intelligence in Healthcare
• Natural Language Processing

Background:

• Integrating heterogeneous healthcare records into a unified application presents significant challenges.
• Ensuring consistent data application across different user bases is complex.

Purpose of the Study:

• To develop MEDSHARE, a web-based application for consolidating patient health records from various sources.
• To implement an NLP-driven diagnostic process within the application.
• To provide patients with accessible health information and disease predictions.

Main Methods:

• Web-based application development for data integration.
• Natural Language Processing (NLP) and fuzzy logic for diagnostic rule generation.
• Support Vector Machine (SVM) classifier for disease prediction.
• Translation packages for multilingual user notifications.

Main Results:

• MEDSHARE successfully integrates diverse health data into a single patient-accessible portal.
• The NLP and fuzzy logic system, coupled with SVM, achieved 89% accuracy in disease prediction.
• Notifications of findings are delivered to users via text message in their native language.

Conclusions:

• MEDSHARE offers a novel solution for healthcare data integration and patient empowerment.
• The application demonstrates the efficacy of NLP and machine learning in clinical decision support.
• Multilingual support enhances accessibility and patient engagement with their health information.