Draft: Round 1 Questionnaire

The following questions are iteratively developed based on factors identified in the content analysis.

1. Please select the most appropriate predictors you believe are essential for developing a predictive model for ART patient retention. (Select all that apply)

• Age

• Sex

• Unemployment status

• CD4 cell count

• Duration of ART treatment

• Adherence to medication regimen

• Presence of comorbidities

• Geographic location

• Others (please specify)

2. In your experience or research, what additional factors do you believe may influence ART patient retention that have not been considered in the above list? Please provide any relevant factors that you think should be included in the predictive model.

3. How would you prioritise the selected predictors in terms of their importance for predicting ART patient retention? Please rank the predictors from most to least important.

4. Can you provide any insights or examples from your own practice or research that support the importance of the predictors you've selected?

5. Please select the most appropriate qualities you believe are essential for a predictive model for ART retention. (Select all that apply).

• Handle missing information

• Handle infrequent/sporadic healthcare events and datasets

• Handle variation in types of elements and dimensions considered

• Handle variations of different health event periods.

• Handle multiple time sequences of data without modelling for separate time sequences converting time sequences into feature vectors.

• Interpretability

6. What challenges or limitations do you foresee in developing a predictive model for ART patient retention based on the selected predictors and predictive model qualities?

7. In your opinion, what methodological considerations should be taken into account when developing and validating the predictive model?

8. Do you have any additional comments or suggestions for improving the development of the predictive model for ART patient retention?