Related Experiment Video
Updated: Jul 7, 2026

A Metadata Extraction Approach for Clinical Case Reports to Enable Advanced Understanding of Biomedical Concepts
Published on: September 20, 2018
Unhealthy alcohol use detection in electronic health records: A comparative study using natural language processing
Xintong Ju1, Jake Solka1, Katherine Weber2
1Department of Psychiatry, University of Michigan, Ann Arbor, MI, USA.
Natural Language Processing (NLP) significantly improves identification of unhealthy alcohol use in primary care compared to structured data (SD). NLP captures more cases, especially those with lower severity alcohol use disorder (AUD), complementing traditional screening.
Area of Science:
- Clinical Informatics
- Public Health
- Digital Health
Background:
- Unhealthy alcohol use, including risky alcohol use and alcohol use disorder (AUD), is often under-identified in primary care.
- Structured data (SD) methods have limitations in capturing the full scope of unhealthy alcohol use.
- Natural Language Processing (NLP) offers a promising avenue for identifying unhealthy alcohol use from clinical notes.
Purpose of the Study:
- To prospectively evaluate the performance of SD and NLP in identifying unhealthy alcohol use in primary care patients.
- To compare the number of positive cases identified by SD and NLP approaches.
Main Methods:
- Extracted electronic health record (EHR) data from a large Midwestern Health System (N=133,144).
- Applied both SD (diagnostic codes, alcohol screening scores) and NLP-based approaches for case identification.
- Recruited 170 participants (85 by SD, 85 by NLP) for gold-standard self-report measures.
Main Results:
- SD identified 820 cases, while NLP identified 48,262 cases of unhealthy alcohol use in the full EHR sample.
- Among SD-identified participants, 41.18% reported AUD and 28.82% reported risky alcohol use.
- Among NLP-identified participants, 20% reported AUD and 27.06% reported risky alcohol use; SD-identified participants had more AUD symptoms and mental health difficulties.
Conclusions:
- NLP substantially expands the identification of unhealthy alcohol use in primary care, capturing cases missed by SD.
- NLP is particularly effective for identifying patients with lower severity alcohol use disorder.
- NLP can serve as a valuable complement to traditional screening methods for comprehensive detection of unhealthy alcohol use.
More Related Videos
Related Concept Videos
Ethers from Alcohols: Alcohol Dehydration and Williamson Ether Synthesis
Ethers can be prepared from organic compounds by various methods. Some of them are discussed below,
Preparation of Ethers by Alcohol Dehydration
In this method, in the presence of protic acids, alcohol dehydrates to produce alkenes and ethers under different conditions. For example, in the presence of sulphuric acid, dehydration of ethanol at 413 K yields ethoxyethane, whereas it yields ethene at 443 K.
Physical Properties of Alcohols and Phenols
Alcohols possess a higher boiling point than aliphatic hydrocarbons of similar...
Alcohols from Carbonyl Compounds: Reduction
Catalytic hydrogenation is similar to the reduction of an alkene or alkyne by adding H2 across the pi bond in the presence of transition metal catalysts like Raney Ni, Pd–C, Pt, or Ru. Aldehydes and ketones can be reduced by this method, often under mild to moderate heat (25–100°C) and...
Protection of Alcohols
Protection
It defines a protecting group as the masking agent to make the more reactive species inert to a given set of conditions. This concept is depicted via the illustration of liquid flow through different outlets in an assembly of pipes. The analogy helps to understand the role...
¹H NMR of Labile Protons: Temporal Resolution
The –OH proton in alcohols typically appears in the range of δ 2 to 5 ppm but can vary depending on the specific...
Mass Spectrometry: Alcohol Fragmentation

