Eye tracking reveals attention patterns in block-based programming. This lets us examine gender-linked differences as students manage the multimodal literacy of coordinating text prompts, color-coded blocks, navigation, and live output. We examined how K–12 students allocate attention while independently completing a short Scratch-like task using a mixed-methods approach. Thirty-two fourth- to sixth-grade students (19 boys, 13 girls) participated. We recorded gaze with a Tobii Spark screen-based tracker and analyzed the data in Tobii Pro Lab. For each student, we computed 187 metrics across four Areas of Interest (AOIs)—Script, Coding Blocks, Output, and Navigation—and analyzed in R 4.3.1 with multiple-comparisons correction. Fifteen metrics differed by gender (seven in Script, four in Coding Blocks, two in Output, two in Navigation). In a qualitative review of eye-tracking videos, girls tended to reread and plan in Script and checked Output later and for longer, whereas boys moved to the workspace sooner, exited Blocks faster with larger, faster eye movements, and made brief early Output checks. Because the study is ongoing, analyses are still in progress and the findings are not yet final; so far, results indicate different strategies, not ability differences.

Gout recurrence is common in real-world settings. While traditional machine learning methods are applicable, their performance is often limited by a lack of diverse data modalities, insufficient understanding of inter-modality interactions, and poor model generalizability. To address these challenges, this work proposes ARL-GRP, a novel framework for forecasting the risk of gout recurrence. This framework is built upon three essential modules: continuous learning utilising real-world multimodel follow-up data, feature representation aggregation employing a pretrained large encoder, and predicting recurrent gout risk using a multilayer perceptron. The experimental comparison demonstrates that our proposed approach generally outperforms conventional machine learning techniques. ARL-GRP can effectively combine structured clinical data and unstructured medical narratives into unified patient representations, significantly outperforming traditional machine learning methods (Accuracy: 0.931, AUC: 0.969). Our method demonstrates strong predictive capability, enabling precise risk assessment and personalised clinical decision-making. Furthermore, the effectiveness of our method is also consolidated through additional analysis using ROC curves and a heatmap.

Testing is an important yet often neglected skill in learning and teaching of computing science at the college level. Prior studies explored integrating test-driven development (TDD) into computer science courses with some degree of success, but also observed issues such as students' lack of appreciation, expressed frustration, and inconsistent adherence to TDD. TDD is a software development methodology that emphasizes writing low-level unit test cases prior to writing the corresponding portion of implementation. Behavior-driven development (BDD) was proposed as an evolution of TDD to emphasize software behavior from users' perspective. BDD has been widely adopted in industry, and holds great potential in addressing the issues in using TDD to improve students' learning of testing. However, BDD was rarely explored in enhancing students' mastery of testing. Informed by the literature, this experience report explored the integration of BDD into a mobile development course. Students' performance, attitude and feedback on BDD was examined, and potential improvement on the integration of BDD was discussed. The results of this report sheds light on how to effectively integrate BDD into computer science courses.

Despite the proliferation of research studies examining elementary computing education worldwide, knowledge regarding the implementation and outcomes of elementary computing education is limited. In this study, we systematically reviewed 45 empirical papers published between January 2015 and December 2021 to establish a holistic understanding of recent research trends, implementation practices, and student outcomes in elementary computing instruction. This review identified three areas of focus that the current literature has explored: 1) curriculum and lesson development; 2) computing tools and learning environment design; and 3) computing pedagogy. In terms of implementation, we identified 1) eight types of subject contexts where computing instruction is implemented; 2) seven types of technologies that are used to engage elementary students in computing; and 3) eight types of learning activities that are implemented. Regarding student outcomes, the current literature reported student outcomes in six dimensions. This review presents the current research landscape of elementary computing education in terms of research trend, implementation practices, and student accomplishment. It showcases the versatility of instructional practices, highlights the progress of current efforts, while also pinpoints potential issues and concerns for future improvement.

Previous research has incorporated physiological data such as heart rate and footsteps to enrich K-12 students' STEM and computing learning experiences. This qualitative study piloted a series of lessons leveraging a novel physiological computing environment with a small group of fifth-grade girls (n=5). The purpose of this study was to understand (1) how the physiological computing lesson activities promoted changes in the students' conceptual understanding of conditional logic and variables and their diverse perspectives in computing and (2) how the students approached problem-solving and what their visual attention looked like during physiological computing. We analyzed multiple sources of data, including students' artifacts, recorded classroom conversations, think-aloud verbalizations, and eye-tracking metrics data. Data analyses revealed that the girls demonstrated an improved understanding of the two computing concepts (i.e., variables and conditional logic), employed different problem-solving strategies, and encountered common challenges such as translating the task instruction to building code with the conditional block. Eye-tracking revealed that the students rarely attended to program output during their programming and debugging processes.

Intelligent virtual agents have tremendous potential for facilitating collaborative learning by modeling and reinforcing desirable collaborative practices. Despite recent work in this area, the extent to which intelligent virtual agents can facilitate improvements in the collaborative behavior of children is largely unknown. This study employed a wizard-of-oz study design and investigated elementary children's collaborative behavior after interacting with virtual agents. These agents model exploratory talk for upper elementary school dyads, such as asking higher-order questions and listening to their partners. The findings uncover associations between elementary learner dyads' positive changes in collaboration after agent interventions, the dyads' affective reactions to interventions, and their attentiveness to the agents. Our results also reveal associations between positive changes in collaboration and the timing of interventions: for example, earlier interventions had a higher occurrence of positive changes, and positive changes in collaboration typically happened within five seconds of interventions. The results suggest ways in which intelligent virtual agents may be used to promote effective collaborative learning practices for children.

Variable is a common computer science (CS) concept and is being introduced to upper elementary students in computational thinking (CT)-integrated instruction. However, there is scant empirical evidence of when and how elementary students should learn variables. For example, national computer science (CS) standards advise introducing variables in grades 3-5 and a K-8 variable learning trajectory (LT) synthesized learning goals from the literature and hypothesized four levels of thinking in working with variables. Yet, little empirical research lies behind these. This mixed methods study examined elementary students' understanding of variables. Participants were sampled from two fourth-grade classes from a Midwestern elementary school that implemented a series of CT-integrated math lessons. Students' written responses to variables assessment items were analyzed. Additionally, cognitive think-aloud interviews were conducted with nine students to elicit students' understanding while solving the variables assessment items. Our findings suggested that most students lacked a conceptual understanding of using variables to create generalized problem solutions that could work with any set of inputs. Additionally, students had difficulty with specific mechanics of using variables such as storing user input in a variable, updating variable values, and using the values stored in variables. This study underscores the need for careful design, use, and analysis of elementary CT-integrated lessons and assessments to introduce and reinforce the conceptual understanding and specific mechanics of variables for elementary students.

Developing computational thinking (CT) assessment methods appropriate for elementary students is attracting growing attention as CT research in elementary education progresses. To review the current elementary CT assessments for potential gaps, and seek additional methodologies to expand our understanding of CT, an integrative literature review of 75 research papers was performed in two phases. In Phase One, we conducted a critical analysis of existing elementary CT assessment studies. Key results include: 1) Artifact analysis, CT assessment items, and interviews are the most common methods utilized to assess CT in elementary grades; 2) Existing CT assessments primarily focus on students' computational artifacts and performance on CT tests; however, strategies to study students' thought processes during CT problem-solving are limited and under-utilized. Guided by the results of phase one, along with the theoretical perspective that connected CT to visual processing ability, in phase two we performed a survey of literature in the area of understanding cognitive processes through eye-tracking (i.e., visual attention) and think-aloud methodologies (i.e., verbalization). We focused on eye-tracking and think-aloud methodologies as these have been used to understand students' cognitive processes during problem-solving in other areas. Based on these findings, we proposed that in addition to current established methodologies, eye-tracking with the think-aloud technique can provide new insights into students' CT.

Elementary schools throughout the United States are attempting to integrate computational thinking (CT) into their instruction, often without guidance from research about effective approaches for achieving particular CT goals. This cross-case study investigated the school-wide integration of CT in three elementary schools in a large urban school district in the Northeast that has a district-led CS for All initiative. Data included interviews with teachers, professional development providers, and school administrators as well as surveys from teachers and classroom observations in each participating school. Findings revealed three distinct approaches to integration: (a) single teacher leader-driven model, (b) scaffolded professional development model, and (c) intensive coaching model. These approaches reflect the visions set by administrators and teachers, methods used by professional development providers, and cultures of each school. Across the case studies, common pedagogical approaches included strategic use of both unplugged and plugged activities with a range of computational tools, a focus on collaborative project-based learning, and the use of CT-specific academic language to anchor new CT learning within the academic disciplines. The study highlighted advantages and challenges within each integration approach with implications for schools considering CT integration.

Debugging has been an expanding topic in K-12 computer science (CS) education research. However, few studies have focused on in-depth analysis of elementary students' debugging in block-based visual programming environments. Thus, using the video analysis technique, this basic interpretive qualitative study aimed to explore what debugging behaviors students exhibited and how these debugging behaviors mapped with an existing K-8 debugging learning trajectory (LT). Findings revealed five types of debugging behaviors and four primary challenges. These debugging behaviors mapped to five consensus goals in the K-8 debugging learning trajectory. Future research will focus on students' efficiency in using debugging strategies and understanding of debugging.

This study aimed to understand elementary students' challenges and interactions in computational thinking-integrated botany through robotics activities. Data was collected from screen-casting videos and analyzed using Collaborative Computing Observation Instrument (CCOI), a web-based analysis instrument with nodes and paths that classify and specify students' computing experience. The results revealed that all participants engaged in independent work for most of the time, with short interactions on 1) general computer technology issues; 2) software navigating issues; 3) questions about academic content; 4) computing discussion with the instructor; 5) informing the instructor about task accomplishment. The findings of this study will provide important insights to CS researchers, educators, and elementary teachers regarding CT-integration research and practice.

For K-8 computer science (CS) education to continue to expand, it is essential that we understand how students develop and demonstrate computational thinking (CT). One approach to gaining this insight is by having students articulate their understanding of CT through cognitive interviews. This study presents findings of a cognitive interview study with 13 fourth-grade students (who had previously engaged in integrated CT and mathematics instruction) working on CT assessment items. The items assessed four CT concepts: sequence, repetition, conditionals, and decomposition. This study analyzed students\textquotesingle articulated understanding of the four CT concepts and the correspondence between that understanding and hypothesized learning trajectories (LTs). We found that 1) all students articulated an understanding of sequence that matched the intermediate level of the Sequence LT; 2) a majority of students\textquotesingle responses demonstrated the level of understanding that the repetition and decomposition items were designed to solicit (8 of 9 responses were correct for repetition and 4 of 6 were correct for decomposition); and 3) less than half of students\textquotesingle responses articulated an understanding of conditionals that was intended by the items (4 of 9 responses were correct). The results also suggested questioning the directional relationships of two statements in the existing Conditionals LT. For example, unlike the LT, this study revealed that students could understand "A conditional connects a condition to an outcome'' before "A condition is something that can be true or false.''

This study investigated school-wide integration of computational thinking (CT) in elementary schools of: 1) systems-level approaches to integration; 2) teachers' understanding and implementation of CT integration, and 3) challenges to integration. Data sources include interviews with teachers, professional development (PD) providers, principals as well as implementation observations. Findings revealed three distinct approaches: (a) Lone STEM teacher implementer, (b) PD scaffolded approach, and (c) Whole school coach-in-residence approach. Teachers generally viewed CT in the context of problem-solving. Although struggles and challenges existed in all three schools, administrators, PD providers, and teachers all had a high commitment to CT integration.