Eating Disorder Dynamic Intervention

Performance of the eating disorder behavior prediction models will be evaluated by computing the area under the receiver operating curve (AUC) for each participants' models on two-weeks of held out (test) data (collected during weeks 5 and 6 of the data collection procedure). The AUC can take on values between 0 and 1, with higher values indicating superior model performance.

Performance of the eating disorder behavior prediction models will be evaluated by computing the Brier score for each participants' models on two-weeks of held out (test) data (collected during weeks 5 and 6 of the data collection procedure). The Brier score can range from 0 to 1 with lower values indicating less model error (i.e., better model performance).

Binge eating occurrence during the data collection protocol will be measured. Binge eating at each ecological momentary assessment survey will be assessed by the question, "Thinking about your most recent eating episode, did you experience loss of control?" (Answer choices = Yes or No).

Dietary restriction will be measured during the data collection protocol. Dietary restriction will be assessed at each ecological momentary assessment survey by the questions, "Thinking about your most recent eating episode, did you limit the amount of food you ate to influence your shape?", "Thinking about your most recent eating episode, did you exclude specific foods to influence your shape or weight?", and Since the last survey, have you skipped a meal or snack?" (Answer Choices = Yes or No).

Dietary restriction will be measured during the data collection protocol. Dietary restriction will be assessed at each ecological momentary assessment survey by the questions, "Since the last survey, have you engaged in any of the following? Self-induced vomiting," "Laxative use (to influence shape or weight)," "Diuretic use (to influence shape or weight)," "Diet pill use (to influence shape or weight)," "Exercise (to influence shape or weight)," and "Other behavior to influence shape or weight" (Answer Choices = Yes or No).

Acceptability will be evaluated via the Feasibility and Acceptability Questionnaire. These data will be collected after participants have completed the 16 weeks of data collection using the EDDI app. Mean ratings ≥ 5 (on the seven-point Likert rating scales) on acceptability items will be considered to indicate acceptability. The response scale is as follows: 1=Strongly Disagree 2=Disagree 3=Disagree a little 4=Neither agree nor disagree 5=Agree a little 6=Agree 7=Strongly Agree, with higher scores indicating greater acceptability.

Acceptability will be evaluated via the System Usability Scale. These data will be collected after participants have completed the 16 weeks of data collection using the EDDI app. Mean ratings ≥ 4 (on the five-point Likert rating scale) on System Usability Scale items will be considered to indicate acceptability. The response scale runs from 1 = Strongly Disagree (minimum value) to 5 = Strongly Agree (maximum value), with higher ratings indicating greater usability. Items 2, 4, 6, 8, and 10 will be reverse coded to align with higher ratings indicating greater usability.

Feasibility of the EDDI app will be evaluated by computing the percentage of eligible participants enrolled. This will be computed by dividing the number of enrolled participants by the total number of eligible participants and multiplying by 100. At least 85% of eligible participants enrolled will be considered to demonstrate feasibility.

Feasibility of the EDDI app will be evaluated by computing the attrition (drop out) rate. This will be computed by dividing the number of participants who discontinue the study prior to the end of study assessment by the total number of enrolled participants and multiplying by 100. An attrition rate of less than 15% will be considered to demonstrate feasibility.

Feasibility of the EDDI app will be evaluated by computing the percentage of JITAIs that produce participant-reported skill use. This will be computed by dividing the instances of reported skills used by the total number of delivered JITAIs and multiplying by 100. JITAIs producing skill use at least 80% of the time will be considered to demonstrate feasibility.

Body image concerns will be measured as a possible mediator of the effect of JITAIs on eating disorder behaviors. Body image concerns will be quantified as the mean rating across the body image items on the Eating Disorder Examination Questionnaire-Short Form, adapted to be rated on a five-point Likert scale as follows: 0=Very slightly/not at all 1 = A little 2 =Moderately 3 = Quite a bit 4= Extremely. Higher mean ratings indicate great body image concerns. Minimum mean score on the items is 0 and maximum is 4.

Cognitive dietary restraint will be measured as a possible mediator of the effects of JITAIs on eating disorder behaviors. Cognitive dietary restraint is assessed by the mean score across four selected dietary restraint items from the Eating Disorder Examination Questionnaire, modified to be assessed on the following response scale: 0=Very slightly/not at all 1 = A little 2 =Moderately 3 = Quite a bit 4= Extremely. The minimum mean score is 0 and the maximum mean score is 4, with higher scores indicating greater dietary restraint.

Negative affect will be measured as a possible mediator of the effects of JITAIs on eating disorder behaviors. Negative affect is assessed by the mean score across six selected negative affect items (upset, hostile, ashamed, afraid, nervous, guilty from the Positive and Negative Affect Schedule, modified to be assessed on the following response scale: 0=Very slightly/not at all 1 = A little 2 =Moderately 3 = Quite a bit 4= Extremely. The minimum mean score is 0 and the maximum mean score is 4, with higher scores indicating greater negative affect.