Authors: Assistant Professor Pradhyumna Yambar, I.Lakshmi Prasanna
Abstract: The rapid usage of smartphones has altered contemporary lives by boosting communication, education, entertainment, and professional activities. In spite of these benefits, smartphone addiction, low productivity, damaged social connections, and other mental and physical health issues are common outcomes of excessive use, which has become a major public health concern. Early identification of people showing addictive smartphone use habits is thus vital for implementing timely prevention measures and individualised therapies. Advances in Machine Learning (ML) have made it possible to analyse behavioural data and accurately identify addiction tendencies using computational methodologies. Using demographic, behavioural, and psychological data gathered from a structured survey, this study introduces a smart machine learning framework for the prediction of smartphone addiction. There are 21 predictive indicators in the dataset that describe things like demographics, smartphone use, and mental health issues including anxiety, depression, and stress. Data is preprocessed by filling in missing values, encoding labels, and normalising features before model building begins in order to maximise learning efficiency. Multiple machine learning algorithms, namely Decision Tree, Random Forest, Logistic Regression, Multi-Layer Perceptron (MLP), and Adaptive Moment Estimation (Adam), are employed to construct predictive models capable of identifying individuals at risk of smartphone addiction. Common performance indicators, like as classification accuracy, are used to assess the trained models. Addiction may be predicted by behavioural variables such as the amount of time spent using a smartphone each day, how often one checks their notifications, how they use applications, their stress levels, their age, and their gender, according to the findings of an experiment. By allowing early addiction identification and encouraging healthy smartphone use patterns, the suggested architecture demonstrates how machine learning may assist healthcare providers, educators, and application developers.
