Psychiatrist.comLeveraging Natural Language Processing to Improve Electronic Health Record Suicide Risk Prediction for Veterans Health Administration Users

Can the fusion of structured and unstructured data reshape the future of suicide risk prediction models?

Novel predictive models that leverage unstructured electronic health record (EHR) data, including clinical notes, can provide significant improvements in suicide risk prediction, according to a recent study.

Key Points:

• The study involved a case-control sample of Veterans Health Administration (VHA) patients in 2017 and 2018, with a focus on patients who died by suicide (n=4,584), each matched with 5 controls who remained alive.
• By employing natural language processing (NLP) on EHR notes, researchers developed predictive models using machine-learning classification algorithms.
• The resulting models demonstrated an overall 19% increase in predictive accuracy, as indicated by the area under the curve (AUC) of 0.69 (95% CI, 0.67, 0.72).
• There was a 6-fold increase in suicide risk concentration for patients at the highest risk tier, relative to the structured EHR model.

Additional Points:

• The benefits of using NLP-supplemented predictive models were evident when compared with conventional structured EHR models.
• The results underscore the potential of integrating structured and unstructured EHR data in suicide risk models in the future.

Conclusion:

• The incorporation of unstructured EHR data via NLP into predictive models significantly enhances suicide risk prediction, indicating the need for a blended approach that utilizes both structured and unstructured data in EHRs.

Psychiatry Further Reading

• Burnout, Depression, and Diminished Well-Being among PhysiciansNovember 20, 2024
• Recommendations to Distinguish Behavioural Variant Frontotemporal Dementia from Psychiatric DisordersNovember 20, 2024
• Why Antidepressant Duloxetine Was RecalledNovember 20, 2024

Did You Know?
According to the Department of Veterans Affairs, veterans are 1.5 times more likely to die by suicide compared to Americans who never served in the military. This underscores the urgent need for improved suicide risk prediction tools within the veteran population.