Research finds that analyzing subtle smartphone movements with AI can predict smoking cravings five minutes in advance

A UK research team analyzed smartphone sensor data from 17 smokers over about 3.5 months.

Using motion data alone, a deep-learning model predicted smoking within 5 minutes, as well as craving and relapse after quitting, with high accuracy.

The study suggests smoking-cessation apps could detect urges in advance from everyday phone movement and deliver personalized interventions.

The work was published in Scientific Reports and used a 1D-CNN-BiLSTM model.