Wallden M, Dahlberg G, Manflod J, et al. Evaluation of 6 years of eHealth data in the alcohol use disorder field indicates improved efficacy of care. Front Digit Health. 2023;5:1282022. doi:10.3389/fdgth.2023.1282022
This article summarizes six years of anonymized data on alcohol relapse, caregiver communication, and novel AI relapse prediction and prevention tools. Previct Alcohol is a Swedish eHealth system used in conjunction with therapy for alcohol use disorder. The system includes daily breathalyzer input, a smartphone application for patients, and a web-based portal for caregivers. Stored data includes digital biomarker analysis for sobriety estimates and relapses. To use the longitudinal data to prevent relapse before it occurs, Previct launched an addition to the app in mid-2021 using AI to predict an oncoming relapse and send an alert to the caretaker. Patient data (breathalyzer, questionnaire results, reported cravings, task completion, etc.) from the previous 1-7 weeks is used to estimate the risk of relapse in the next 1-3 days. The study analyzed data from 269,913 patient days before and 106,018 patient days after the introduction of the relapse prediction tool. Overall, the use of Previct protected users from the spike in relapse seen in other research. In addition, the duration of relapses was shorter, and the number of relapses was lower than expected based on the pre-predictor data following the introduction of the relapse predictor tool. While specifics were not recorded, there is a consistent increase in caregiver-patient communication within the Previct platform after an alert is sent. This suggests that when the predicted relapse alerts are sent, caretakers intervene and prevent the relapse, or reduce the duration of the relapse. While the data anonymity prevents some of the desired analytics continued use and analysis of Previct Alcohol’s relapse prediction tool are promising.