Pain-level detection is vital to determine proper medical treatment. Existing self-reporting, behavioral, and image-based pain detection methods typically lead to high costs for professional staff and clinical equipment and also have a high risk of leaking sensitive information, which could be exploited by adversaries to learn gender, age group, and underlying health conditions. To address these challenges, we propose a privacy-aware voice-based pain-level detection framework, which extracts frequency-domain characteristics from audio files, removes the gender-sensitive features from the identified characteristics, and trains a convolutional neural network (CNN) to perform 3-level pain classification. Our preliminary analysis and experiments based on a commonly adopted database (TAME Pain dataset) indicate that log-Mel spectrogram and zero-crossing rate are two promising voice features for effective pain-level classification. We further extend our study to a nonverbal dataset, VIVAE (Variably Intense Vocalizations of Affect and Emotion Corpus), and confirm that the identified voice features are applicable to both verbal and nonverbal patients for pain classification.

Real-time recognition and localization of keyboard keys combined with real-time hand tracking and detection of finger-press actions are critical functions of vision-based virtual keyboards. These are not trivial tasks because there are numerous types of keyboard layouts and typist behaviors. Furthermore, key detection is especially difficult in uncontrolled real-world scenarios, where users' hands occlude significant portions of the keyboard while typing. Also, the detection finger press-down actions is further complicated in-the-wild by frequently changing camera viewpoints without direct line of sight to the finger pressing a key. In this work, we address the challenge of complete keyboard segmentation and detection of hand-occluded keys by proposing a deep learning approach for realtime finger-press detection, visible-key detection, and occluded-key recovery. Our models were trained and evaluated on Kaggle's Keyboard Key Detection dataset [1] and further empirically tested on the MSU Typing Behavior Database [3]. Our best models achieved 88% finger press-down detection and a key detection performance of 0.91 IOU and 97.8% mAP@75, while running at 60 fps.

Metacognition is widely acknowledged as a key soft skill in collaborative software development. The ability to plan, monitor, and reflect on cognitive and team processes is crucial to the efficient and effective functioning of a software team. To explore students' use of reflection--one aspect of metacognition--in undergraduate team software projects, we analyzed the online chat channels of teams participating in agile software development projects in two undergraduate courses that took place exclusively online (n = 23 teams, 117 students, and 4,915 chat messages). Teams' online chats were dominated by discussions of work completed and to be done; just two percent of all chat messages showed evidence of reflection. A follow-up analysis of chat vignettes centered around reflection messages (n = 63) indicates that three-fourths of the those messages were prompted by a course requirement; just 14% arose organically within the context of teams' ongoing project work. Based on our findings, we identify opportunities for computing educators to increase, through pedagogical and technological interventions, teams' use of reflection in team software projects.