JMIR Publications

ABSTRACT

Background:

Wearable sensors, decision-support systems, and digital biomarkers increasingly generate data that physicians must appraise and act on. Medical curricula rarely offer dedicated training in these technologies; where digital health content exists, it is typically elective, postgraduate, or theoretical, without contact with real sensor data. Published reports also rarely provide the details needed to reproduce a hands-on course or to anticipate where students struggle.

Objective:

This tutorial describes how we designed, delivered, and evaluated a mandatory preclinical digital biomarkers course for second-year medical students at ETH Zürich, CH. Using the second edition of the course as a use case, we share the course structure, the sensor pool, and our evaluation instrument, and report students’ self-reported learning gain and the tasks they found most difficult.

Methods:

The course applies a studio-based learning approach, combining foundational lectures with group projects in which students receive a commercial wearable sensor, formulate a health-related research question, design a measurement approach, collect and manage the resulting data, and present potential clinical applications. Participation in data collection was voluntary and assessment was ungraded. At course completion, students were invited to complete an anonymous voluntary survey covering perceptions of the course, prior experience, perceived learning gain, workload, challenges encountered, support needed, and large language model (LLM) use. Responses were summarized descriptively; associations were explored using one-way ANOVA, Fisher exact test, Spearman rank correlation, and linear regression, with two-sided significance set at P<.05.

Results:

Thirty-three students responded. Prior experience ranged from none (30.3%) to university- or research-related (9.1%). Students spent a mean of 1.6 (SD 1.1) hours/week testing devices and 1.3 (SD 0.6) hours/week working with the data. Learning gain was rated rather strong by 63.6% and very strong by 6.1%, with open-text responses centering on awareness of the op-portunities, limitations, and reliability of wearable sensing. Data-related tasks were the prin-cipal obstacle: 72.7% of students reported difficulty with data access, management, or inter-pretation, accounting for 58.3% of all reported challenges, against 10.4% for device setup. Most students used LLMs, predominantly to understand how the devices worked. All anal-yses were exploratory and no statistically significant associations were detected.

Conclusions:

A mandatory, hands-on digital biomarkers course is feasible within the preclinical curriculum, and most respondents judged it to deliver a substantial learning gain. Data handling, not device operation or clinical content, was the rate-limiting step, indicating that applied digital health teaching at this stage requires explicit scaffolding in data competencies. We provide our sensor specifications and evaluation instruments so that other programs can adapt the format. Clinical Trial: N/A