A Feasibility Study of the Kasi Learning System to Support Independent Use of STEM Diagrams by Students With Visual Impairments

Abstract

Introduction: Visual model comprehension and application are important for success in science, technology, engineering, and mathematics (STEM) courses. As educational materials shift to primarily digital content with dynamic interactive visuals, students with visual impairments are at risk for being disadvantaged, since few interactives are born accessible. To fill this gap of accessible digital STEM learning tools, we designed and tested the Kasi Learning System. Kasi uses tactile manipulatives and computer vision with audio-based augmented reality algorithms to provide a multisensory experience of an interactive digital image. Methods: Ten high school students who are visually impaired (ie, blind or have low vision) participated in an underpowered random control study to evaluate the feasibility and usability of Kasi by completing an active learning lesson. The control group was instructed by a human, whereas the Kasi group was instructed by a computer. Follow-up interviews with both students and their instructors provided further insight. Results: Comparing the experiences of the two groups suggests that Kasi is an effective instructor for completing the activity. Comparison of students who chose to use braille versus large-print pieces revealed that braille users found the system to be easier to use. Discussion: All students efficiently identified the pieces. Regarding the audio, students who do not typically use a screen reader repeated the prompts more frequently and took longer to adapt to the system. Those in the Kasi group demonstrated increased engagement as shown by the increase in submitted answers. Overall, Kasi users’ performance improved significantly during the lesson. Implications for Practitioners: Kasi is most readily adapted and used by those who do not rely on vision. However, students with low vision may benefit from using a tool like Kasi earlier in their schooling to strengthen their auditory and tactile skills. Kasi appears to have the potential to provide students independence in studying STEM diagrams.

The Lewis Structure Explorer: Accessible by Design

“…33 It was previously reported that one such UDL-designed interface for chemical particulate concepts, which used tactile manipulatives recognized by computer vision algorithms, provided an accessible input method for the chemistry interactives in a K-12 educational setting. 34 Though useful for foundational spatial concepts, this multisensory interface is limited by the need for external hardware. A digital-only interface was desired to overcome these limitations and to provide a more scalable solution for building additional accessible digital interactives.…”

“…An approach for creating inclusive accessibility solutions is outlined with the Universal Design for Learning (UDL) framework to eliminate systemic barriers to success by using multiple means for “engagement, representation, and action” thus matching the widest possible range of users’ needs . It was previously reported that one such UDL-designed interface for chemical particulate concepts, which used tactile manipulatives recognized by computer vision algorithms, provided an accessible input method for the chemistry interactives in a K-12 educational setting . Though useful for foundational spatial concepts, this multisensory interface is limited by the need for external hardware.…”

“…Due to the low-incidence of the BLV population, which number less than 0.5% of student population in K-12, and is even a lower incidence population in college-level chemistry due to the BLV students steering away from STEM courses, group-based research studies for establishing the promise of outcomes for a learning tool with BLV students are not feasible. Even within this low-incidence population, there are significant differences in spatial perception between individuals with low-vision, or those whose blindness is either congenital or happened later in life . Single case design (SCD) methods, where BLV individuals serve as their own controls and the introduction of an intervention (the learning tool) is staggered across a small sample of participants, are the most common method for building the evidence base to demonstrate the effectiveness of a learning intervention with this population. − Two SCD studies are planned with K-12 BLV students using accessible interfaces.…”