AI helps clarify complex array and interaction of risk factors contributing to alcohol use disorder
Explainable AI (xAI) models are illuminating the multiple risk factors that underlie the development of alcohol use disorder (AUD), according to a new study. Although some factors involved in AUD—including demographic, socioeconomic and genetic variables—are known, researchers typically study them in isolation, and their roles in individuals' risk are often small. The variability of AUD…
Recent research published in Alcohol: Clinical & Experimental Research reveals how artificial intelligence (AI) can help uncover the complex interplay of risk factors contributing to alcohol use disorder (AUD). By leveraging advanced explainable AI (xAI) techniques, scientists have developed a model that analyzes a vast array of variables to identify those most relevant to AUD development.
This comprehensive study examined data from 12,178 participants aged 40–69, comprising case and control groups based on alcohol use evaluations and diagnoses, sourced from the UK Biobank.
The xAI model incorporated over 400 factors encompassing five key domains: demographic, socioeconomic, mental health unrelated to substance use, nonalcoholic substance use, and biological factors, including genetic combinations and brain structure. The model's analysis pinpointed demographic, substance use, and biological elements as the most influential in predicting AUD risk.
Sex, lifetime tobacco smoking, cannabis use, and age emerged as particularly significant predictors. Other critical factors included genetic ancestry, socioeconomic status, mental health status, regional brain structure, and genetic susceptibility to AUD.
However, the study also uncovered that some variables, such as socioeconomic status and mental health issues, did not substantially enhance the model's predictive capabilities. This finding suggests that these factors may have overlapping influences, leading to redundancy in their predictive power. Remarkably, the research demonstrated nonlinear interactions between biological and environmental risk factors, emphasizing the complex ways in which these variables influence AUD development.
Notable interactions highlighted include the combined impact of sex and age, as well as the relationship between social isolation and a specific component of ancestry.
These insights not only underscore the potential of AI in identifying and elucidating the multifaceted risk factors of AUD but also highlight the necessity for researchers to carefully select relevant variables and consider their interactions. While the study's findings offer valuable insights into the predictors of AUD, they may not be universally applicable, particularly to non-European populations due to potential differences in genetic and environmental factors.
The researchers advocate for a judicious approach in selecting risk factors and acknowledging their complex interactions to improve the predictive accuracy of AI models in understanding and addressing AUD.
Written by urgent.news from Medical Xpress's reporting — not their text. Machine-written — may contain errors; check the original before relying on it.
