Abstract
AIMS: This study was conducted to verify the effectiveness of using digital self-care devices in reducing alcohol-related problems among high-risk alcohol users in community addiction-related institutions. METHODS: Data were collected from 257 adults in Korea aged 18 and over (157 men and 60 women), examining their level of alcohol use disorder and the usage of digital self-care devices (such as the number of days alcohol consumption was logged, continuous days of sobriety, feeling, alcohol cravings, alcohol probability, etc.). RESULTS: The results confirmed that the severity of alcohol use disorder significantly decreased before and after the use of digital self-care devices, as analyzed by a t-test (M = 5.239, SD = 10.121, t = 6.945, df = 179, P = .000***). Additionally, a machine learning analysis (random forest) was conducted to explore the factors that most influence the reduction in alcohol risk levels among participants. The analysis revealed that the factor "continuous days of sobriety" had the most significant impact on the reduction of alcohol risk levels. The predictive accuracy of this factor was demonstrated using an ROC curve (AUC = 0.724). Subsequently, a multiple regression analysis was conducted to explore the factors influencing continuous days of sobriety. The results indicated that age and the logging of sobriety days had a significant impact, with the logging of sobriety days emerging as the most influential factor. CONCLUSION: These results suggest that in reducing alcohol consumption and achieving successful sobriety, it may be more important to maintain continuous sobriety rather than the total number of sober days. Additionally, it is necessary to identify the key factors that help maintain continuous sobriety. Understanding which elements need to be fulfilled through digital self-care devices to sustain continuous sobriety is also essential. CLINICAL TRIAL REGISTRATION: https://cris.nih.go.kr/cris/search/detailSearch.do?search_lang=EM&focus=reset_12&search_page=M&pageSize=10M&page=undefinedM&seq=16267M&status=5M&seq_group=16267, identifier, KCT0005135.