A Systematic Temporal Extraction Pipeline for Medical Concepts in Clinical Notes

Area of Science:

• Medical Informatics
• Natural Language Processing
• Clinical Data Analysis

Background:

• Natural Language Processing (NLP) is increasingly used to extract clinical features from electronic health records (EHRs).
• Temporal information is crucial for interpreting medical concepts and their changes over time within clinical notes.
• Understanding longitudinal changes in patient conditions is vital for effective healthcare.

Purpose of the Study:

• To introduce a systematic framework, NLP annotation-Relaxation-Generation (NRG), for analyzing temporal changes in medical concepts from EHRs.
• To demonstrate the utility of the NRG pipeline in tracking changes in specific clinical features for patients with inflammatory bowel disease.

Main Methods:

• The NRG framework compiles medical concept changes by integrating status annotations and timestamps from multiple clinical notes.
• The pipeline was applied to analyze changes in "extra-intestinal manifestations" and "medications" for inflammatory bowel disease patients.

Main Results:

• The NRG pipeline successfully identified and compiled incidents of medical concept changes over time.
• The framework demonstrated its ability to provide insights into the longitudinal evolution of clinical features.
• Effectiveness was shown in tracking changes for specific concepts in inflammatory bowel disease cohorts.

Conclusions:

• The NLP annotation-Relaxation-Generation (NRG) framework is effective for analyzing temporal dynamics of medical concepts in EHRs.
• NRG offers valuable insights into longitudinal changes in clinical features at both individual and population levels.
• This approach enhances the understanding of disease progression and treatment patterns over time.