A Clinical Decision Support Framework for Heterogeneous Data Sources.

Area of Science:

• Medical Informatics
• Health Data Management
• Clinical Decision Support Systems

Background:

• Increasingly complex and diverse health data requires advanced management solutions.
• Existing systems struggle to integrate heterogeneous data from various sources.
• Effective utilization of electronic health records (EHR) is crucial for improved patient care.

Purpose of the Study:

• To propose a novel clinical decision support (CDS) framework for integrating and managing heterogeneous health data.
• To develop a method for disease prediction and complication identification using integrated patient data.
• To enhance physician diagnostic and treatment efficiency through data-driven recommendations.

Main Methods:

• Developed a CDS framework to consolidate diverse health data (lab results, patient info, health records).
• Employed multilabel classification on integrated EHR data for disease recommendation.
• Improved the k-nearest neighbors algorithm (CML-kNN) to leverage correlations among diseases for enhanced prediction.

Main Results:

• The proposed CDS framework effectively integrates heterogeneous health data.
• Multilabel classification accurately recommends potential diseases to physicians.
• The CML-kNN algorithm demonstrates improved performance in identifying related diseases and complications.
• Experimental results validate the framework's effectiveness and practicality on real-world health data.

Conclusions:

• The developed CDS framework offers a practical solution for managing complex health data.
• The integration of data and advanced algorithms aids in efficient disease diagnosis and complication prediction.
• This approach has the potential to significantly assist physicians in clinical decision-making and patient management.