Restricted natural language based querying of clinical databases

Area of Science:

• Clinical Informatics
• Data Science
• Natural Language Processing

Background:

• Healthcare professionals require efficient tools to extract knowledge from complex clinical data.
• Existing methods for clinical data analysis can be cumbersome and require specialized query language expertise.

Purpose of the Study:

• To develop a generic translation algorithm for converting restricted natural language queries (RNLQ) into SQL.
• To provide a user-friendly interface for healthcare professionals to access and analyze clinical data.

Main Methods:

• Introduction of CliniDAL, a clinical data analytics language with six questioning templates.
• Development of a translation algorithm using a similarity-based Top-k algorithm for mapping RNLQ to SQL.
• Implementation of a two-layer rule-based method for interpreting temporal expressions in queries.

Main Results:

• CliniDAL's interface allows easy RNLQ composition with over 84% accuracy in mapping query terms to data models.
• Temporal expressions in RNLQ are accurately mapped to clinical information system time entities.
• The system supports various data design models, including Entity-Relationship (ER) and Entity-Attribute-Value (EAV).

Conclusions:

• CliniDAL offers a simplified mechanism for knowledge extraction from diverse clinical information systems.
• The integration of generic mapping, translation algorithms, and a temporal analyzer enhances data analytics capabilities.
• The tool empowers healthcare professionals to leverage clinical data more effectively for improved patient care.