Common Laboratory Parameters Are Useful for Screening for Alcohol Use Disorder: Designing a Predictive Model Using...

Area of Science:

• Biochemistry
• Data Science
• Clinical Diagnostics

Background:

• Diagnosing alcohol use disorder (AUD) is challenging, leading to underdiagnosis in some patients.
• Biochemical markers and data science offer potential for improved AUD detection.
• Early identification of AUD is crucial for preventing associated medical complications.

Purpose of the Study:

• To identify an optimal combination of laboratory markers for detecting alcohol consumption using data science.
• To develop and validate predictive models for alcohol consumption risk.
• To enhance the accuracy of alcohol use disorder diagnosis.

Main Methods:

• An analytical observational study involving 337 participants (204 with AUD, 133 controls).
• Collected medical history and laboratory markers from a hospital database.
• Developed and compared three predictive models: logistic regression, classification tree, and Naive Bayes network, utilizing Python's scikit-learn and Weka.

Main Results:

• The Naive Bayes network model achieved the highest prediction accuracy at 87.46%.
• Logistic regression yielded a maximum accuracy of 85.07%.
• Classification tree offered simpler interpretation with 79.4% accuracy.

Conclusions:

• Combining common biochemical markers with data science significantly enhances the detection of alcohol use disorder.
• The developed models show promise in identifying individuals at risk of alcohol consumption.
• Improved AUD detection can lead to timely interventions and prevention of future health issues.