Enhance Everyday Devices to Better Understand and Positively Influence Our Behavior

What is it about?

With the proliferation of ubiquitous devices that have passive sensing and active interaction capabilities, there's a unique opportunity to understand and support user well-being. The research focuses on detecting risky behaviors and addressing health concerns using human-centered applied ML to understand passive behavior logs over time. The main challenges addressed are: 1. Achieving interpretable, personalized, and generalizable behavior modeling techniques. 2. Using the models to provide intelligent, adaptive intervention experiences to users for better well-being. The goal is to create deployable, intelligent interventions for health and well-being that utilize ML-based behavior models. The paper emphasizes the importance of devices understanding human behavior and impacting it to support various aspects of human well-being. The research aims to bridge modeling techniques and intelligent intervention designs to better support users' behavior goals.

Featured Image

Photo by Emma Simpson on Unsplash

Photo by Emma Simpson on Unsplash

Why is it important?

It addresses the need for a more human-centric approach in AI and ML applications, especially in the domain of health and well-being. The paper highlights the challenges in behavior modeling, such as the need for interpretability, personalization, and generalizability. It underscores the importance of not just sensing and modeling behavior but also effectively translating these models into intelligent systems that can positively impact users. The research aims to provide a transparent, trustful, and deployable system that can be tailored to individual nuances, ensuring predictions and interventions are personalized. Furthermore, the paper discusses the importance of generalizability across datasets for real-world deployment.

Perspectives