Diabetes Engineering Studios Show How Student Analysis Shifts From Concepts to Quantitative Models

Biomedical engineering students working through diabetes-focused design challenges applied a structured engineering analysis model at every course level, but their work differed in ways that tracked the deliberately different expectations of sophomore and junior courses, according to a new study published in Biomedical Engineering Education. The research, led by Stephanie Fuchs, Mridusmita Saikia, and Jonathan Butcher at Cornell University, analyzed final design presentations from 18 student teams to examine how the PRIME studio model—Physiological system identification, Ranked constraints, Interactive block diagrams, Mathematical modeling, and Evidence-based solutions—shapes student reasoning as the same clinical problem is revisited at two points in the curriculum.

The study addresses what the researchers call the inspiration gap in biomedical engineering education: the disconnect between abstract engineering principles and their meaningful application to healthcare challenges. Engineering curricula often emphasize theoretical content while offering limited opportunities for authentic problem-solving, and biomedical engineering has historically lacked consensus on the foundational knowledge all graduates should possess, unlike more established disciplines with a recognizable core of shared content. The PRIME studio model was introduced across all studio-integrated courses of Cornell’s redesigned BME curriculum. In each three-hour studio session, teams of four to five students work through a structured sequence: they identify the physiological systems and dysfunctions involved in a challenge, collaboratively develop and prioritize engineering constraints, create interactive block diagrams to model system relationships, examine mathematical relationships among system variables at a level appropriate to their preparation, and finally propose evidence-based solutions informed by their analyses.

Diabetes serves as the instructional anchor for this approach. As a high-burden-of-care disease with significant complications and ongoing therapeutic limitations, it presents students with interconnected layers of biology, physiology, materials science, human factors, and engineering that must be systematically integrated. Because studio-integrated courses span the BME curriculum from sophomore to senior year, students repeatedly revisit the same disease through different technical lenses. The researchers analyzed final design presentations from two complementary challenges: a sophomore-level artificial pancreas design task in a human physiology course, and a junior-level organ-on-a-chip modeling task in a cellular systems course, both completed by a cohort of approximately 70 students progressing through the program in 2024.

The assignments were intentionally differentiated according to students’ curricular preparation. Sophomore students, who had not yet completed the relevant mathematics sequence, emphasized conceptual and qualitative systems analysis: they identified physiological dysfunctions in Type 1 Diabetes, evaluated existing artificial pancreas technologies using ranked constraints, and constructed system-level block diagrams to conceptualize design improvements. Junior students, having completed the mathematics sequence, were explicitly required to formulate differential equations representing glucose homeostasis dynamics, develop in silico microfluidic schematics for a multi-organ pancreas–liver–muscle model, and generate computational modeling results supporting their proposed solutions. The researchers stress that course-level comparisons therefore characterize student work produced under developmentally scaffolded implementations rather than performance under identical assignment conditions.

To evaluate the presentations, the team coded student work across three dimensions—constraint development, block-diagram complexity, and solution-analysis alignment—using the PRISM rubric, which aligns directly with the PRIME model and assesses both whether a principle was applied and how effectively. Coding was performed with high inter-rater reliability, measured by Gwet’s AC1 statistics ranging from 0.733 to 1.000, a statistic chosen for its stability with skewed distributions. Mann–Whitney U tests compared ordinal rubric scores between course levels, with Bonferroni-adjusted p values reported for the block-diagram and solution-alignment comparison families. Six of fourteen sophomore teams and twelve of seventeen junior teams had selected diabetes-related design review projects, yielding eighteen team presentations for analysis.

The constraint results were among the clearest in the study. Sophomore teams identified 66 constraints across six teams, while junior teams identified 87 across twelve teams, and 15 of the 18 teams developed explicit priority-ranking systems. But quality diverged sharply. Junior teams produced significantly more specific constraints, with 82 percent of junior constraints receiving the highest specificity score compared with just 14 percent of sophomore constraints. Sophomore constraints clustered around user-centered themes—cost, safety, ease of use, size, and portability—reflecting the patient-facing artificial pancreas challenge, whereas 60 percent of junior constraints were technical and engineering-oriented, involving cell viability, glucose and insulin dynamics, fluid flow, and model reproducibility, consistent with the organ-on-chip challenge’s emphasis on physiological modeling. The theme distributions differed significantly between course levels, with a large effect size.

Block-diagram performance showed selective course-level differences rather than uniform improvement. Junior teams scored significantly higher on variables and constants, component interactions, and systems thinking, with large effect sizes. Their schematics frequently included specific cell-line designations such as INS-1, HepG2, C2C12, and RPTEC, defined variables, and engineering notation illustrating how pancreatic, hepatic, and muscle compartments function as integrated networks through shared flow and biochemical signaling pathways. Junior presentations more consistently represented multi-organ integration and dynamic responses to biochemical signals. However, two mathematical-content metrics told a more complicated story. Equations and formulas did not differ significantly between course levels, and the junior median of zero indicates that at least half of the junior presentations contained no mathematical expressions, despite the explicit requirement. Quantitative parameters were generally absent from presentations at both levels.

An intriguing descriptive pattern emerged in feedback mechanism representation: sophomore teams actually scored higher than junior teams, though the difference was not statistically significant after correction. Sophomore students, approaching glucose regulation from a direct physiological perspective, were more likely to incorporate explicit feedback loops showing how cells and organs respond to changes in glucose and insulin concentrations. Junior teams, focused on structural interactions between cell chambers within their organ-on-chip devices, predominantly represented static compartmentalization rather than dynamic cellular responses to physiological fluctuations. The researchers interpret this as a warning: as students advance toward technically detailed modeling challenges, explicit attention to feedback mechanisms must be maintained, since these dynamic responses remain fundamental to understanding therapeutic potential even within static organ-on-chip systems.

Solution-analysis alignment revealed the study’s largest between-course difference in Quantitative Support, which was expected given the intentionally different assignment requirements. Junior teams received uniformly high quantitative support scores, with several teams developing solutions that functioned as rigorous quantitative investigations—one proposed pulsatile insulin delivery at 15-minute intervals with dosing proportional to measured glucose concentration, using simulation results to evaluate whether the intervention maintained glucose within the target range. Yet constraint-solution and diagram-solution alignment did not differ significantly between course levels, with both groups earning intermediate scores, suggesting that teams did not always make the relationships among their analytical steps explicit. The researchers describe the intended process as constraint-driven creativity: solutions informed and bounded by relevant biological and engineering requirements, with mathematical modeling positioned as a tool for understanding and constraining the problem space rather than an isolated academic exercise.

The authors acknowledge important limitations. The analysis focused on team deliverables rather than individual work, team compositions changed between courses, and the inaugural implementation of PRIME studios involved ongoing pedagogical troubleshooting, including a considerably refined PRISM rubric by the junior course. The different prompts—artificial pancreas design versus organ-on-chip modeling—introduce task variability that may confound direct comparisons, and the limited sample restricts statistical power. Nevertheless, the findings offer the BME education community a scalable instructional approach for bridging clinical inspiration with analytical rigor. For sophomore students, the results suggest value in developing quantifiable constraints that function as design checklists; for junior students, they identify continued need for measurable constraints and explicit feedback mechanism representation. With future cohorts, the team plans to investigate how targeted refinements influence developmental patterns, positioning PRIME as a pathway for meaningful integration of clinical inspiration with systematic engineering analysis across the biomedical engineering curriculum.

Subject of Research: Evaluation of student engineering analysis in scaffolded diabetes-focused biomedical engineering studio courses

Article Title: From Device Design to Systems Modeling: Evaluating Student Development in Diabetes Engineering Analysis

Article References: Fuchs, S., Saikia, M., & Butcher, J. (2026). From Device Design to Systems Modeling: Evaluating Student Development in Diabetes Engineering Analysis. Biomedical Engineering Education. https://doi.org/10.1007/s43683-026-00249-2

Image Credits: AI Generated

DOI: 10.1007/s43683-026-00249-2

Keywords: biomedical engineering education, PRIME studio model, diabetes, Type 1 Diabetes, artificial pancreas, organ-on-a-chip, systems modeling, constraint development, block diagrams, curriculum design, Cornell University, engineering education