The goal of this study is to develop a machine-learning guided recovery messaging system. The main question it aims to answer is can messages be used to: * help people to improve their health * make changes in people's lives to address alcohol and substance use Participants will: * complete surveys * use a recovery-support digital therapeutic app
This study seeks to optimize messaging components which can be implemented in a recovery support messaging system such as may accompany a digital therapeutic app, in order to determine optimal messaging to increase interaction with recovery support resources, and whether messaging has any effect on clinical outcomes.
Study Type
INTERVENTIONAL
Allocation
NA
Purpose
TREATMENT
Masking
NONE
Enrollment
416
Automated recovery support messaging system for participants with alcohol use disorder (AUD), paired with a machine learning guided relapse risk prediction model.
University of Wisconsin
Madison, Wisconsin, United States
RECRUITINGParticipant engagement with the STAR system
Measured as days of engagement with the recovery system.
Time frame: data collected at 2 and 4 months
Number of heavy drinking days in the previous 28 days
Data collected via Timeline Followback (TLFB) conducted by interview at 0, 2, and 4 months. Ecological Momentary Assessment (EMA) lapse reports will be used to supplement the TLFB. These are daily surveys where the participant answers the question "have you had a drink you have not yet reported?" (Yes/No). If yes they are asked to indicate the date of their drinking and how many drinks they consumed that day (less than 1, 1-2, 3, 4, 5-6, 7-8, over 8). Four or more drinks are considered heavy drinking days for women and greater than 4 drinks are considered heavy drinking days for men.
Time frame: data collected at baseline, 2, 4 months
Number of drinking days in the previous 28 days
Data collected via Timeline Followback (TLFB) conducted by interview at 0, 2, and 4 months. EMA lapse reports will be used to supplement the TLFB. These are daily surveys where the participant answers the question "have you had a drink you have not yet reported?" (Yes/No). If yes, they are asked to indicate the date of their drinking.
Time frame: data collected at baseline, 2, 4 months
Flourishing Scale Score
10 items rated on 0-10 Likert scale where lower scores represent worse functioning and higher scores represent better functioning. Total possible range of scores is from 0-100.
Time frame: data collected at baseline, 2, 4 months
Generalized Anxiety Disorder-7 (GAD-7) Score
7 items rated on 0-3 Likert scale where lower scores represent better functioning and higher scores represent worse functioning. Total possible range of scores is from 0-21.
This platform is for informational purposes only and does not constitute medical advice. Always consult a qualified healthcare professional.
Time frame: data collected at baseline, 2, 4 months
Multidimensional Inventory of Recovery Capital (MIRC) Score: Total Score
1-4 Likert scale. Some items are reverse coded but higher total scores represent better functioning and lower total scores represent worse functioning. Scored as total score 4-56.
Time frame: data collected at baseline, 2, 4 months
Multidimensional Inventory of Recovery Capital (MIRC) Score: Each Domain
1-4 Likert scale. Some items are reverse coded but higher total scores represent better functioning and lower total scores represent worse functioning. Scored in 4 domains (social, physical, human, and cultural), each from 1-14.
Time frame: data collected at baseline, 2, 4 months
Patient Health Questionnaire (PQH-9) Score
9 items rated on 1-4 Likert scale. Total possible range of scores from 0-27 where lower scores represent better functioning and higher scores represent worse functioning.
Time frame: data collected at baseline, 2, 4 months
Digital Working Alliance Inventory (D-WAI)
Six items rated on 1-7 Likert scale. Scored in 3 domains (Goals, Tasks, Bonds, each with 2 items) with possible scores from 2 to 14 where higher scores represent greater alliance with the STAR system in that domain.
Time frame: data collected at 2 and 4 months
Message Usefulness
Seven items rated on 1-7 Likert scale. Total possible scores from 7 to 49 where higher scores represent greater valuation of STAR message usefulness.
Time frame: data collected at 2 and 4 months
Trust of the System
Ten items rated on 1-7 Likert scale. Total possible scores from 10 to 70 where higher scores represent greater trust in the STAR system.
Time frame: data collected at 2 and 4 months