Table 4.

Predictors of change in cigarettes per day after 6 months, in decreasing order of importance

Predictors measured at baseline or after 1 month	%IncMSE
Cigarettes/day as measured at baseline	46.11
Use of nicotine medications after one month	12.49
Use of e-cigarettes as self-reported after one month	8.25
Intention to quit smoking as measured after one month	7.65
Number of different days when the app was opened (automatically collected)	6.53
Number of times the app was opened (automatically collected)	6.39
Minutes before the first cigarette of the day	4.27
Current use of any smoking cessation app after one month	3.91
App’s use duration (automatically collected)	3.85
Perceived helpfulness of the app after one month	3.71
Age	3.01
Used of heated tobacco at baseline	3.01
Number of years smoking	2.98
Used of heated tobacco product at one month	2.69
Sex	2.33
Smoking other tobacco products at baseline	1.22
Use of nicotine medications at baseline	1.13
Use of e-cigarettes at baseline	1.05
Being depressed	0.23
Country (Switzerland vs. France)	0.09
Experiment group	0.02

%InMSE = per cent increase in mean squared error, a statistical measure indicating the level on the predictor variable importance in the regression machine learning algorithm