Personalized Behavioral Risk Prediction Technology
Unmet Need
Daily environments significantly influence mental and physical health, impacting factors like stress, sleep quality, and behaviors tied to addiction and mental health conditions. For example, specific environments may trigger smoking, exacerbate insomnia, or contribute to depressive episodes. Identifying and modifying these spaces to reduce negative influences can play a critical role in improving overall wellness.
With over 90% of the US population owning a smartphone, mobile technologies offer a scalable, cost-effective way to assess environmental risks and suggest actionable changes. Unlike traditional approaches, these technologies can leverage real-time environmental analysis to enable just-in-time adaptive interventions (JITAIs), providing users with timely and personalized recommendations to enhance wellbeing. There is a growing need for systems that can integrate into mobile health (mHealth) applications, empowering individuals to better understand their environments, mitigate behavioral triggers, and promote healthier outcomes through tailored feedback.
Technology
Duke inventors have developed a computer vision system that predicts behavioral risks based on images of users' environments. This system can be integrated into digital therapeutic applications to help individuals identify high-risk settings, such as those that may trigger stress, smoking, or insomnia, and suggest modifications to reduce these risks.
The system combines a camera and computing device to analyze user-uploaded images using a trained convolutional neural network. It identifies objects and environmental features associated with increased risks, such as a cluttered workspace that hinders focus or a location tied to smoking habits. The technology provides real-time feedback, allowing users to make changes to their surroundings that promote healthier behavior.
This system has been validated in a study focusing on environments associated with smoking, analyzing over 4,900 images. It accurately predicted high-risk smoking environments with an AUC of 0.827, demonstrating performance comparable to smoking cessation experts.
Other Applications
Beyond smoking cessation, this technology can support:
- Mental health: Identifying environments that may contribute to anxiety, depression, or insomnia, with recommendations to create more calming spaces.
- Addiction management: Helping users pinpoint and avoid triggers for alcohol consumption, overeating, or other addictive behaviors.
- General wellness: Offering environmental suggestions to improve sleep, focus, and relaxation through design and lighting changes.
- Asthma and allergies: Detecting triggers such as dust or allergens to prevent attacks.
This system can also inform behavioral science research and architectural designs to create more supportive environments for mental and physical health.
Advantages
- Real-time detection: Automatically identifies environmental triggers and risks, enabling timely and personalized interventions.
- Seamless integration: Easily integrates into existing mobile health apps, wearable devices, and sensors.
- Cost-effective and accessible: Utilizes smartphone cameras for continuous monitoring, making it more affordable than traditional interventions.
- Broad adaptability: Applicable to a variety of health behaviors and conditions, from stress reduction to smoking cessation.
- Empowers users: Provides actionable insights to help individuals create healthier, more supportive environments tailored to their needs.