IRB 倫理審查通過摘要

1. Basic information approved by IRB

The internal reference number of the IRB protocol corresponding to this summary is v3.0 (finalised on 16 March 2026), and it was approved on 10 April 2026. When this centre first made a public disclosure, it used v1.0 as the starting version number for the public disclosure version, in line with the versioning system for governance documents.

2. Application motivation and Umbrella design

The Principal Investigator is the founder and administrator of the Uedu platform, which has already accumulated more than 3,000 learners, over 255,000 AI dialogue records, and coverage across 567 course classes. As the number of collaborating instructors continues to grow, it is necessary to establish a unified research ethics framework so that all research activities are conducted on the premise of safeguarding participants' rights and interests.

This Umbrella IRB has three purposes:

1. Systematic evaluation of the potential risks to participants posed by various research activities on the platform
2. To provide partner teachers with common ethical rules and standard operating procedures, and to ensure that teachers across institutions follow consistent informed consent, de-identification and data management standards
3. Protect participant rights at the system level through platform technical measures, such as the automatic de-identification export in Uedu Lab, rather than relying only on the self-discipline of individual researchers

3. Three-tier classification of research scope

This Umbrella IRB divides all research activities on the platform into three tiers, A, B, and C, clearly defining the scope of IRB coverage for each tier:

4. Research activity categories

The research activities covered by this Umbrella IRB are divided into two categories: “retrospective research” and “newly collected research”. The overall scale is the platform’s cumulative total of over 3,000 learners and 4,000+ active users; the specific scale of each individual research activity is set out in the full protocol.

4.1 Retrospective study (secondary analysis of existing platform data)

• AI dialogue history analysis: classification of student–AI dialogue messages by Bloom's Taxonomy cognitive levels
• Peer discussion analysis: discussion messages from the Forum and GitHub Discussion
• Platform usage data: learning traces, submission times, dwell times, etc.

4.2 New research collection (individual consent)

• Wearable physiological sensing: obtain HRV, sleep, stress and other indicators via Garmin Connect, Apple HealthKit and Google Health Connect (opt-in, no limit on number of participants)
• UeduBrain multimodal brain–neuro sensing experiment: EEG + fNIRS + PPG synchronous sensing (Category B, PI only)
• Gender diversity and gender identity survey: questionnaire under the SGBA Plus framework
• Semi-structured in-depth interview: qualitative interviews with gender-diverse learners (Category B, PI only)
• Expert focus group: experts in educational technology and learning analytics
• Learning traits exploration survey: Holland RIASEC, Big Five, OEJTS/MBTI
• IoT environmental sensing: classroom-level light, temperature and humidity, noise, CO₂ (non-personal data)
• Screen / Webcam Classroom Recording: digital classroom observation on the Instructor side and Learning history recording on the Student side

The specific recruitment scale, implementation setting, number of returns and other execution details for each activity are set out in the full protocol; this summary does not disclose specific figures so as to preserve flexibility for future implementation.

5. Informed consent mechanism

There are six consent forms for this project, signed according to the category of research activity:

5.1 Consent handling for retrospective data

For secondary analysis of existing retrospective platform data, this is handled in accordance with the ethics committee's 'waiver of consent for retrospective data' standard: the exported data have been automatically de-identified through Uedu Lab, and the data were originally collected as records of the normal operation of teaching services. The de-identified dataset used for research cannot be reversed to identify individuals.

6. Data categories and retention period

The categories of data processed in this project include:

• AI conversation and LLM interaction records
• Learning behaviour record (clicks, dwell time, submission trail)
• Derived analysis data (Bloom cognitive level, learning trait scale scores, language complexity indicators)
• Physiological data from wearable devices (HRV, sleep, stress indicators)
• Brain and neural sensing signals such as EEG, fNIRS and PPG (Category B)
• In-depth interview recordings and verbatim transcripts (Category B)
• Screen / Webcam / microphone recordings (mp4 / wav)
• Gender identity survey responses
• IoT environmental sensing data (classroom level, non-personal)

6.1 Dual-track retention strategy

The above retention policy is consistent with the Privacy and data retention policy: platform-side data is retained permanently; self-service account deletion currently only deactivates the account, and deletion of personal data is handled case by case upon email request; external exports by researchers are limited to 5 years, and exported data does not include names, email addresses, or student ID numbers. See Privacy and data retention policy for details.

7. Deletion and destruction procedure

This project’s data management SOP (IRB protocol appendix D) specifies the following processing workflow:

1. Collection phase: conducted within the scope disclosed in the consent form; data are labelled with categories and the relevant research activity at the time of collection
2. Processing phase: raw data are stored on access-controlled servers or laboratory-dedicated storage; personal data are stored in encrypted fields
3. Research use phase: use de-identified datasets exported via Uedu Lab for analysis; re-linking pseudonyms to real identities is not permitted
4. Destruction phase：
   • Platform-side data: when a user exercises the right to erasure, it is handled under §3 of the Privacy and data retention policy (self-service deletion deactivates the account; full deletion is handled case by case upon email request)
   • Non-platform data: retained for 5 years after this project is closed; upon expiry, it shall be manually destroyed under the PI's direct supervision (permanent file deletion, shredding of hard-copy documents, and formatting and overwriting of storage media)

8. Ethics safeguards

• De-identification: Research export data automatically removes direct identifiers such as name, student number, email and IP address
• Encrypted transmission: all external connections use TLS
• Anti-Outing: data related to gender identity and diverse genders is hidden by default in the teacher interface; only the PI may access a de-identified version for research purposes
• Access control: researcher access requires PI authorisation; access records are retained for audit
• Ethics training: all co-researchers may participate only after completing NTU-REC or equivalent ethics training
• Withdrawal mechanism: the participant may withdraw consent at any time; withdrawal does not affect academic results or course entitlements; after withdrawal, the data will be deleted or continue to be processed (de-identified version) according to their request
• Considerations regarding unequal teacher-student relationships: Research data are marked with research codes and are not linked to course grades during analysis; the PI does not use participation in the research as a basis for grading

9. Application for a complete Protocol

This summary discloses only key structured information for preliminary review by international collaborators and their institutions' data protection officers and ethics review bodies. The full IRB protocol (including the main text, 6 informed consent forms, and 5 appendices covering the research data inventory/research instruments/interview outline/data management SOP/recruitment materials, totalling over 200 pages) is provided via the following procedure:

Full approval letter and complete IRB protocol available upon written request to [email protected], subject to a Researcher Access Agreement.

Please explain the following in your email:

• Applicant name, institution, contact details
• Purpose of application (collaborative research, ethics review, academic reference, etc.)
• Whether they are willing to sign an NDA
• Whether you already have an approval letter from the IRB of your institution (if this involves collaborative research)

The platform reserves the right to decide whether to provide access according to the purpose of the application; for institutional DPO or IRB review requirements, a response will in principle be provided within 30 days of receiving complete information.

10. Case number verifiability

The validity and basic information of case number NTU REC 202507EM058 may be directly verified with the Research Ethics Committee of National Taiwan University College of Medicine. This platform does not publicly provide scanned copies of the approval letter for the following reasons: (i) the approval letter contains personally identifying information about the principal investigator; (ii) the case number itself is independently verifiable and does not need to be duplicated by a scanned copy; (iii) the disclosure practice of international research institutions (such as Stanford, ETH and UZH) also follows a “summary + original on request” model.

11. Relationship with other governance documents

12. Contact point

This summary is a public disclosure document; details of specific research activities, recruitment scale, compensation amounts and so on are not disclosed in this summary. Please refer to the full protocol.