HappyKidz: Privacy Preserving Phone Usage Tracking

We propose a smartphone app named HappyKidz that allows parents to monitor their child’s well-being in a non-invasive way based on measurable behavioral indicators. The app collects behavioral data on smartphone usage, encrypts them with homomorphic encryption, and sends the encrypted data to a server. The server calculates a well-being score for the child using a trained neural network and sends the resulting encrypted score to the parent to decrypt locally. This architecture takes advantage of modern machine learning techniques while maintaining privacy for individual children from the server. Unlike existing apps, it does not directly control, access or report raw behavioral data. In this work, we describe the high-level application and implement a proof of concept of the core neural network logic. We address concerns about the appropriate use of the app and discuss potential barriers to implementation, including collecting appropriate training data and scaling the model to a larger feature set.