Sunday 09 March 2025
Researchers have developed a new model that can accurately predict the risk of developing cannabis use disorder (CUD) in young people. The study, published in a recent issue of a scientific journal, used data from over 10,000 individuals to train and test the model.
The model takes into account various factors that are known to increase the likelihood of CUD, such as a family history of substance abuse, early use of cannabis, and mental health issues. It also considers personality traits like neuroticism, conscientiousness, and openness to experience.
To develop the model, the researchers used a combination of statistical techniques and machine learning algorithms. They first identified the most important risk factors for CUD using a technique called Cox proportional hazards modeling. This allowed them to determine which variables were most strongly associated with the development of CUD.
Next, they used a type of machine learning algorithm called Bayesian elastic net to combine these risk factors into a single model. This approach helped to identify the most relevant predictors and reduce the risk of overfitting, where a model becomes too specialized to the training data and fails to generalize well to new cases.
The researchers then tested their model using data from two separate studies: the Add Health study and the Christchurch Health and Development Study (CHDS). The Add Health study followed over 10,000 individuals from adolescence into young adulthood, while the CHDS study tracked a cohort of over 1,000 children from birth to age 30.
In both datasets, the model performed well, accurately predicting the risk of CUD in young people. For example, it correctly identified more than 70% of individuals who went on to develop CUD within five years of starting cannabis use.
The researchers also found that the model was more accurate when used to predict CUD in individuals with a family history of substance abuse or those who started using cannabis at an early age. This suggests that the model may be particularly useful for identifying high-risk groups and targeting interventions accordingly.
Overall, this study demonstrates the potential of machine learning algorithms to improve our understanding of CUD and develop more effective prevention strategies. By combining multiple risk factors into a single model, researchers can gain a more complete picture of the complex relationships between cannabis use and mental health outcomes.
The findings also highlight the importance of considering individual differences in personality traits and early life experiences when predicting the risk of CUD.
Cite this article: “New Model Predicts Risk of Cannabis Use Disorder in Young People”, The Science Archive, 2025.
Cannabis Use Disorder, Machine Learning, Young People, Risk Prediction, Substance Abuse, Mental Health, Family History, Early Cannabis Use, Personality Traits, Bayesian Elastic Net
