Table 2 The survey questions for this study were adapted from the ‘substance disorder screener’ of the Global Appraisal of Individual Needs–Short Screener (GAIN-SS) questionnaire6, with question numbers and abbreviations added for this study for substance use disorder (SUD) prediction

From: Predicting substance use behaviors with machine learning using small sets of judgment and contextual variables

Question #

Abbreviation

Question

In the past one month

2–3 months ago

4–12 months ago

1+ year ago

Never

4

3

2

1

0

1

alcohol

You used alcohol weekly or more often?

     

2

cannabis

You used cannabis weekly or more often?

     

3

opioid

You used heroin, fentanyl, or other opiates?

     

4

stimulant

You used a stimulant like cocaine or meth?

     

5

time spent

You spent a lot of time either getting alcohol or other drugs, using alcohol or other drugs, or recovering from the effects of alcohol or other drugs (e.g., feeling sick)?

     

6

social problems

You kept using alcohol or other drugs even though it was causing social problems, leading to fights, or getting you into trouble with other people?

     

7

isolation

Your use of alcohol or other drugs caused you to give up or reduce your involvement in activities at work, school, home or social events?

     

8

withdrawal symptoms

You had withdrawal problems from alcohol or other drugs like shaky hands, throwing up, having trouble sitting still or sleeping, or you used any alcohol or other drugs to stop being sick or avoid withdrawal problems?

     
  1. The first four questions are referred to ‘substance use’ variables, and the last four questions are referred to ‘SUD behavior’ variables. Please note that the four questions relating to substance use are an expansion of the one question about substance use framed in the original GAIN-SS.
  2. Participants responded to when was the last time they experienced the following things two or more times based on five-time blocks: 0 = Never; 1 = 1+ years ago; 2 = 4–12 months ago; 3 = 2–3 months ago; 4 = In the past month. For recency prediction analysis, the responses from each question were divided into two classes recent and not-recent, where recent included the responses from the past one year (responses: 2, 3, and 4) and not-recent responses included never and more than a year ago (responses: 0 and 1).
  3. For the total score computation, one or more responses from the four substance use variables from the past one year were counted only once, and the responses from the past one year for SUD behavior variables were counted and added to the composite score. The total GAIN-SS scores ranged from 0 to 5 and are referred to as ‘composite severity’. For composite severity prediction analysis, total scores were divided into high and low classes based on threshold values of 1–5. All values below a given threshold were labeled as ‘low’, and values above and equal to the threshold were labeled as ‘high’.