Extraction of Geriatric Syndromes from Discharge Summaries: Development of a Novel Dataset

Overview

This study presents a novel annotated dataset for extracting geriatric syndromes (GS) from hospital discharge summaries using natural language processing (NLP).

Background

Geriatric syndromes, such as falls, delirium, and dementia, significantly impact the quality of life in older adults and often involve multiple organ systems. Accurate identification of these syndromes is crucial for improving clinical care and outcomes. However, they are frequently underrepresented in structured electronic health records, necessitating innovative approaches like NLP for extraction from unstructured clinical text.

Data Highlights

Key Findings

• BERT-cased achieved the highest F1-score of 0.897 for document-level labelling.
• BioClinicalBERT performed best for negation considerations with an F1-score of 0.888.
• Frailty, falls, and delirium were the GS entities with the highest F1-scores (1.0, 0.973, and 0.946, respectively).
• NER-C task results indicated that context-aware labels for falls and frailty were more accurately identified when implied rather than explicitly stated.
• Document-level aggregation reduced inconsistencies in model performance.

Clinical Implications

The development of a detailed annotation framework for geriatric syndromes can enhance the accuracy of NLP models in clinical settings. Understanding the characteristics that influence model performance can guide future research and implementation of NLP tools in healthcare.

Conclusion

This study underscores the potential of NLP to improve the extraction of geriatric syndromes from clinical texts, emphasizing the need for high-quality annotated datasets to enhance model effectiveness.

Related Resources & Content

1. American Journal of Epidemiology, 2026 -- Automated Detection of Fall-Associated Injuries in Unstructured Clinical Documentation
2. Frontiers in Medicine, 2026 -- Rapid clinical identification of vulnerable older adults in acute and emergency care: narrative review of short geriatric screening tools
3. Frontiers in Digital Health, 2026 -- Harvesting and Analyzing Intensive Care Chart Information from Patient Data Management Systems
4. Psychiatry.org, 2025 -- American Psychiatric Association Publishes Updated Comprehensive Guideline for the Prevention and Treatment of Delirium
5. NICE, 2025 -- Falls: assessment and prevention in older people and in people 50 and over at higher risk
6. Frontiers in Medicine — Burden of geriatric and cognitive disorders and the impact of integrated care models on morbidity, functional decline, and health service utilization among older adults
7. New Clinical Practice Guideline for Blood-Based Biomarkers
8. ESPEN Guideline on Clinical Nutrition and Hydration in Geriatrics
9. Psychiatry.org - American Psychiatric Association Publishes Updated Comprehensive Guideline for the Prevention and T
10. Overview | Falls: assessment and prevention in older people and in people 50 and over at higher risk | Guidance | NICE
11. Effectiveness of drug interventions to prevent delirium after surgery for older adults: systematic review and network meta-analysis of randomised controlled trials | The BMJ
12. From Cerebrospinal Fluid to Blood Draw: Plasma p-Tau217 as a Non-Invasive Biomarker for Alzheimer's Disease: A Fagan Nomogram-Based Meta-Analytic Study - PubMed
13. Usefulness of the global leadership initiative on malnutrition -GLIM- criteria to identify malnutrition in older adults: systematic review and meta-analyses - PubMed
14. Extracting geriatric syndromes from electronic health records: a scoping review - PMC