New paper accepted to ACM #IMWUT / @ubicomp: Toward Proactive Support for Older Adults — Predicting the Right Moment for Providing Mobile Safety Help. The work is led by the lab’s student Tamir Mendel, with Roei Schuster, Eran Tromer, and Eran Toch. 
We predict whether a user might need help from friends and family when interacting at a certain moment with a mobile device. This is geared towards creating proactive social support technology for older adults, identifying the moments where they might need support. To collect data, we have asked users of various ages (n=150) to indicate whether they would like to receive help when interacting with various scenarios, including scenarios that required them to manage privacy settings etc.
We then used the information to build a prediction model, with random forest models slightly outperforming other types of models. We have used SHAP to interpret the model, and to analyze the scenarios and the user characteristics that may be used to tailor proactive support.
Image