Evaluating the effectiveness of AI-enabled Digital Detox Tools for screen time management, stress reduction, and quality of life among IT Professionals

Abstract

As the digital landscape grows more reliant on technology, extended screen time, techno-stress and work–life imbalance has become major concerns for IT workers, necessitating the need for effective digital well-being interventions with AI. This paper conducts a multidimensional computational comparative analysis of the effectiveness of twelve commercially available AI-based digital detox tools based on a common framework. A multi-modal dataset containing 200 IT professionals was combined with a publicly available dataset, and used to populate a Python-based analytical pipeline, which included behavioural data, physiology data from wearables, psychometric measures, and qualitative data. In the proposed framework, systematic preprocessing, feature engineering, descriptive statistics, paired sample t-test, effect size analysis, participant clustering, and comparative performance ranking were applied. To create a composite measure for objective comparison of the different intervention scenarios, a Unified Digital Well-Being Profile (UDWP) was developed that incorporates behavioural, physiological, psychometric and qualitative indicators.
The computational assessment revealed statistically significant modelled improvements in several aspects of digital wellbeing. Estimated reductions in perceived stress were 10.19%, an 11.44% reduction in daily screen time and 9.44% reduction in after-hours digital usage, with an increase in work–life balance of 18.22%. Other improvements in sleep duration, heart-rate variability and qualitative sentiment were noted. Significant differences between baseline and simulated post-intervention outcomes were identified through statistical analysis with large practical effects for reduction in screen-time and improvement in sentiment. Headspace was simulated to have the best multidimensional performance among the scenarios evaluated in the proposed computational evaluation framework. The proposed UDWP aims to establish a repeatable computational model for evaluating AI-based digital well-being technologies, a process that incorporates disparate evidence into a standard process of evaluation. The results provide concrete recommendations for organisations, occupational health professionals and technologists to choose digital detox solutions by learning from AI and offer a clear and replicable method to support future research comparing digital detox solutions in the workplace.