Authors: Temkar Sakshi Sandip, Bangar Abhishek Popat, Bhosale Sachin Bajirao, Khatri Anand A.
Abstract: Smartphones, personal computers and tablets are deeply embedded in education, work and social life. Alongside the benefits, unregulated screen use is associated with compulsive checking, disturbed sleep, reduced sustained attention and psychological distress, a pattern commonly described as digital addiction or problematic device use. Screen-time features built into mobile operating systems report totals and enforce fixed limits, but they do not distinguish productive from compulsive use, do not anticipate the onset of a compulsive episode, and rely on rigid blocking that users often bypass. A structured narrative review was carried out on screen-time tools, behavioural signals of compulsive use, ma-chine learning (ML) methods for usage modelling, persuasive and nudging interventions, and privacy-preserving de-ployment. The evidence shows that simple telemetry such as unlock frequency, app-switching speed, session structure and late-night use carries predictive signal, that no single signal uniquely indicates addiction, and that most reported models are trained on small samples, validated without user-grouped or temporal splits, and never tested in long-term field use. The review contributes (i) a comparative analysis of existing screen-time solutions, (ii) a behavioural-signal evidence table, (iii) an eight-layer, literature-justified conceptual architecture for an on-device monitoring and interven-tion assistant with uncertainty handling and a gated validation loop, (iv) a leakage-resistant evaluation protocol, and (v) a prioritised agenda of research gaps that must be closed before such systems can be deployed with confidence.
