Caterina Fuster-Barcelo, Gonzalo R. Rios-Munoz, Arrate Munoz-Barrutia
This study examines the integration of digital collaborative tools and structured peer evaluation in the Machine Learning for Health master's program, through the redesign of a Biomedical Image Processing course over two academic years. The pedagogical framework combines real-time programming with Google Colab, experiment tracking and reporting via Weights & Biases, and rubric-guided peer assessment to foster student engagement, transparency, and fair evaluation. Compared to a pre-intervention cohort, the two implementation years showed increased grade dispersion and higher entropy in final project scores, suggesting improved differentiation and fairness in assessment. The survey results further indicate greater student engagement with the subject and their own learning process. These findings highlight the potential of integrating tool-supported collaboration and structured evaluation mechanisms to enhance both learning outcomes and equity in STEM education.