Automated classification of encounter notes in a computer based medical record

Area of Science:

• Medical Informatics
• Natural Language Processing
• Pediatric Asthma Research

Background:

• Computerized medical record systems are valuable but underutilized for textual data.
• Manual review of clinical notes is time-consuming and resource-intensive.
• Automated analysis of electronic health records (EHRs) offers potential efficiency gains.

Purpose of the Study:

• To evaluate the efficacy of automated information systems in analyzing clinical notes.
• To determine the extent to which automated systems can replace manual chart review.
• To identify acute exacerbations in pediatric asthmatics using text classification of EHRs.

Main Methods:

• Application of INQUERY, a probabilistic inference net information retrieval system.
• Utilizing FIGLEAF, an inductive decision tree text classifier.
• Classification of electronic encounter notes for pediatric asthma exacerbations.

Main Results:

• Both INQUERY and FIGLEAF achieved average precisions exceeding 80%.
• An enhanced version of INQUERY with relevance feedback performed as the top-performing system.
• Automated classification demonstrated high accuracy in identifying specific clinical events.

Conclusions:

• Automated information retrieval and text classification systems show significant potential for analyzing clinical text.
• These systems can effectively identify acute exacerbations in pediatric asthmatics from encounter notes.
• Further refinement and integration of these technologies could enhance medical record analysis and reduce manual review burden.