Search for: All records

There is a growing body of research revealing that longitudinal passive sensing data from smartphones and wearable devices can capture daily behavior signals for human behavior modeling, such as depression detection. Most prior studies build and evaluate machine learning models using data collected from a single population. However, to ensure that a behavior model can work for a larger group of users, its generalizability needs to be verified on multiple datasets from different populations. We present the first work evaluating cross-dataset generalizability of longitudinal behavior models, using depression detection as an application. We collect multiple longitudinal passive mobile sensing datasets with over 500 users from two institutes over a two-year span, leading to four institute-year datasets. Using the datasets, we closely re-implement and evaluated nine prior depression detection algorithms. Our experiment reveals the lack of model generalizability of these methods. We also implement eight recently popular domain generalization algorithms from the machine learning community. Our results indicate that these methods also do not generalize well on our datasets, with barely any advantage over the naive baseline of guessing the majority. We then present two new algorithms with better generalizability. Our new algorithm, Reorder, significantly and consistently outperforms existing methods on most cross-dataset generalization setups. However, the overall advantage is incremental and still has great room for improvement. Our analysis reveals that the individual differences (both within and between populations) may play the most important role in the cross-dataset generalization challenge. Finally, we provide an open-source benchmark platform GLOBEM- short for Generalization of Longitudinal BEhavior Modeling - to consolidate all 19 algorithms. GLOBEM can support researchers in using, developing, and evaluating different longitudinal behavior modeling methods. We call for researchers' attention to model generalizability evaluation for future longitudinal human behavior modeling studies. 

Feeling a sense of belonging is a central human motivation that has consequences for mental health and well-being, yet surprisingly little research has examined how belonging shapes mental health among young adults. In three data sets from two universities (exploratory study: N = 157; Confirmatory Study 1: N = 121; Confirmatory Study 2: n = 188 in winter term, n = 172 in spring term), we found that lower levels of daily-assessed feelings of belonging early and across the academic term predicted higher depressive symptoms at the end of the term. Furthermore, these relationships held when models controlled for baseline depressive symptoms, sense of social fit, and other social factors (loneliness and frequency of social interactions). These results highlight the relationship between feelings of belonging and depressive symptoms over and above other social factors. This work underscores the importance of daily-assessed feelings of belonging in predicting subsequent depressive symptoms and has implications for early detection and mental health interventions among young adults. 

The NSF-funded Redshirt in Engineering Consortium was formed in 2016 with the goal of enhancing the ability of academically talented but underprepared students coming from low-income backgrounds to successfully graduate with engineering degrees. The Consortium takes its name from the practice of redshirting in college athletics, with the idea of providing an extra year and support to help promising engineering students complete a bachelor’s degree. The Consortium builds on the success of three existing “academic redshirt” programs and expands the model to three new schools. The Existing Redshirt Institutions (ERIs) help mentor and train the new Student Success Partners (SSP), and SSPs contribute their unique expertise to help ERIs improve existing redshirt programs. This Work in Progress paper describes the history of the Redshirt in Engineering Consortium; the Redshirt model as a framework for addressing issues related to diversity, equity, and inclusion in engineering; and initial lessons learned from the implementation of the model across unique institutional contexts. 

Low-income students are underrepresented in engineering and are more likely to struggle in engineering programs. Such students may be academically talented and perform well in high school, but may have relatively weak academic preparation for college compared to students who attended better-resourced schools. Four-year engineering and computer science curricula are designed for students who are calculus-ready, but many students who are eager to become engineers or computer scientists need additional time and support to succeed. The NSF-funded Redshirt in Engineering Consortium was formed in 2016 as a collaborative effort to build on the success of three existing “academic redshirt” programs and expand the model to three new schools. The Consortium takes its name from the practice of redshirting in college athletics, with the idea of providing an extra year and support to promising engineering students from low-income backgrounds. The goal of the program is to enhance the students’ ability to successfully graduate with engineering or computer science degrees. This Work in Progress paper describes the redshirt programs at each of the six Consortium institutions, providing a variety of models for how an extra preparatory year or other intensive academic preparatory programs can be accommodated. This paper will pay particular attention to the ways that institutional context shapes the implementation of the redshirt model. For instance, what do the redshirt admissions and selection processes look like at schools with direct-to-college admissions versus schools with post-general education admissions? What substantive elements of the first-year curriculum are consistent across the consortium? Where variation in curriculum occurs, what are the institutional factors that produce this variation? How does the redshirt program fit with other pre-existing academic support services on campus, and what impact does this have on the redshirt program’s areas of focus? Program elements covered include first-year curricula, pre-matriculation summer programs, academic advising and support services, admissions and selection processes, and financial aid. Ongoing assessment efforts and research designed to investigate how the various redshirt models influence faculty and student experiences will be described. 

The NSF-funded Redshirt in Engineering Consortium was formed in 2016 with the goal of enhancing the ability of academically talented but underprepared students coming from low-income backgrounds to successfully graduate with engineering degrees. The Consortium takes its name from the practice of redshirting in college athletics, with the idea of providing an extra year and support to help promising engineering students complete a bachelor’s degree. The Consortium builds on the success of three existing “academic redshirt” programs and expands the model to three new schools. The Existing Redshirt Institutions (ERIs) help mentor and train the new Student Success Partners (SSP), and SSPs contribute their unique expertise to help ERIs improve existing redshirt programs. The redshirt model is comprised of seven main programmatic components aimed at improving the engagement, retention, and graduation of students underrepresented in engineering. These components include: “intrusive” academic advising and support services, an intensive first-year academic curriculum, community-building (including pre-matriculation summer programs), career awareness and vision, faculty mentorship, NSF S-STEM scholarships, and second-year support. Successful implementation of these activities is intended to produce two main long-term outcomes: a six-year graduation rate of 60%-75% for redshirt students, and increased rates of enrollment and graduation of Pell-eligible, URM, and women students in engineering at participating universities. In the first year of the grant (AY 16-17), SSPs developed their own redshirt programs, hired and trained staff, and got their programs off the ground. ERIs implemented faculty mentorship programs and expanded support to redshirt students into their sophomore year. In the second year (AY 17-18), redshirt programs were expanded at the ERIs while SSPs welcomed their first cohorts of redshirt students. This Work in Progress paper describes the redshirt programs at each of the six Consortium institutions, identifying distinctions between them in addition to highlighting common elements. First-year assessment results are presented for the ERIs based on student surveys, performance, and retention outcomes. Ongoing research into faculty experiences is investigating how participation as mentors for redshirt students changes faculty mindsets and instructional practices. Ongoing research into student experiences is investigating how the varied curricula, advising, and cohort models used across the six institutions influence student retention and sense of identity as engineering students.