Related Experiment Video
Updated: Jun 13, 2026
07:31
Implementation of a Real-Time Psychosis Risk Detection and Alerting System Based on Electronic Health Records using CogStack
Published on: May 15, 2020
Developing a Natural Language Processing Strategy to Avoid Biased Data in Electronic Health Record Suicide Risk
Maxwell Levis1,2, Monica Dimambro1, Joshua Levy3
1VAMC White River Junction White River Junction Vermont USA.
Psychiatric Research and Clinical Practice
|June 12, 2026
Summary
This study uses natural language processing to accurately identify electronic health record (EHR) data before patient death, reducing bias in suicide risk models.
Area of Science:
- Medical Informatics
- Computational Linguistics
- Public Health
Background:
- Electronic health records (EHRs) are valuable for suicide risk modeling.
- Inaccurate death dates in EHRs can bias suicide prediction models.
- Previous methods excluded significant pre-death EHR data.
Purpose of the Study:
- To improve the accuracy of detecting EHR reported death dates.
- To reduce bias in suicide risk prediction models by better utilizing pre-death EHR data.
- To investigate natural language processing (NLP) for precise EHR data validation.
Main Methods:
- Utilized NLP to analyze EHR texts from Veterans Affairs patients who died by suicide.
- Developed code to classify texts as pre- or post-death/suicidal action.
- Applied this approach to a corpus of 9127 EHR texts.
Main Results:
- Identified 274 texts entered after the reported death date.
- Identified 1556 texts chronologically after death indicators but not explicitly dated.
- The NLP approach retained 60.9% of interval data, significantly more than prior exclusion methods.
Conclusions:
- The NLP method enhances the detection of valid pre-mortem EHR data.
- This approach minimizes bias in suicide risk modeling.
- Improved risk prediction can strengthen suicide prevention services.
Related Concept Videos
01:29Bias in Epidemiological Studies
Biases can arise at various stages of research, from study design and data collection to analysis and interpretation. Recognizing and addressing these biases is essential to ensure the validity and reliability of epidemiological findings.Broadly speaking, biases in epidemiology fall into three main categories: selection bias, information bias, and confounding. A more detailed description of possible biases is:
01:25Strategies for Assessing and Addressing Confounding
Confounding is a critical issue in epidemiological studies, often leading to misleading conclusions about associations between exposures and outcomes. It occurs when the relationship between the exposure and the outcome is mixed with the effects of other factors that influence the outcome. Given that, addressing confounding is of high importance for drawing accurate inferences in research.
Confounding can be addressed at both the design phase of a study and through analytical methods after data...
Confounding can be addressed at both the design phase of a study and through analytical methods after data...