Research roadmap overview

Educational Omics — a six-dimensional framework analogous to bio-omics, using AI, physiological sensing and classroom practice to systematically map learning experiences, and now deployed across 24 universities in Taiwan.

24 Cross-campus deployment

4,600+ Registered users

420K+ AI conversation messages

23 Research papers

5 semesters in operation

Educational Omics six-dimensional framework

Analogous to bio-omics (Genomics, Proteomics,...), it systematically understands learning experience across six analytical dimensions. Each dimension can operate independently — schools without wearables can still use Cognomics and Sociomics.

Cognomics

Research Subject: Cognitive Processes

Automatically classify Bloom's cognitive levels (six levels) from AI dialogue traces to track Student cognitive development. Knowledge-graph extraction captures concept mastery and establishes cross-institutional, cross-disciplinary Bloom's norm benchmarks.

Linguomics

Research Subject: Language Expression

Use Whisper speech-to-text and semantic embedding vector analysis to record classroom audio and screen recordings. The "Instructor speech → automatic Worksheets" workflow transforms classroom content into learning materials in real time.

PhysioNeuromics

Research Subject: Physiological and Neural Signals

National Science and Technology Council funded project

Integrate consumer wearable devices (Garmin, Apple Watch, Health Connect). PALM (Physiological-Aware Language Model) adaptively adjusts AI teaching dialogue strategies according to the learner's physiological baseline.

Sociomics

Research Subject: Social Interaction

Forum, collaborative editing (Yjs CRDT), live polling, anonymous question wall. Combines validated psychometric scales (Holland RIASEC, Big Five, OEJTS) for learning trait analysis.

Environomics

Research Subject: Learning Environment

Integrate open government environmental data — weather stations from the Central Weather Administration and air quality stations from the Ministry of Environment — to match the weather and air-quality exposure at the course location. Seat-level IoT edge sensing (light, temperature and humidity, noise, CO₂) and classroom spatial visual sensing (attendance, group behaviour analysis) are planned extensions.

Ethicomics

Research Subject: Ethical Governance

Manage fine-grained consent mechanisms at each dimension and IRB research ethics review. Students can control data visibility, maintain a complete access audit trail, and export data in compliance with GDPR requirements.

Platform sub-systems

Six modular subsystems, each implementing data collection and analysis for a dimension of Educational Omics

Uedu Core

AI TA, quizzes, surveys, forums

Uedu Fit

Garmin / Apple Watch / Health Connect

Uedu Mind

PALM Physiologically-Aware Adaptive Teaching

Uedu Sense

Government open environmental data (CWA / Ministry of Environment), with reserved expansion for IoT edge sensing

Uedu Brain

EEG + fNIRS + PPG (in development)

Uedu Lab

Research data export interface

Funded research project

Supported by a competitive project from the National Science and Technology Council (NSTC), driving the core development of the PhysioNeuromics dimension within the Educational Omics research framework.

Protocol number

NSTC 115-2410-H-008-028

Approved Period of Implementation

2026.08 – 2027.07

Principal Investigator

Assistant Professor Chia-Kai Chang

Implementing body

Center for General Education, National Central University

This project develops a Physiologically-Aware Language Model (PALM), based on Affective Computing, integrating multimodal Learning history data such as conversational semantics, operational behaviour and physiological signals to explore the dynamic mechanisms of human–machine collaborative learning, and thereby to construct an information education theory model. The project centres on the Uedu platform developed by the research team, adopting a dual-track strategy of 'solid foundation + autonomous extension': the foundational layer uses the integrated Uedu Fit (Garmin HRV) combined with the PAD emotion model to infer learners' operational states, and adaptively adjusts AI tutoring dialogue strategies; the extension layer independently develops Uedu Brain, a low-cost classroom physiological sensing device integrating EEG, fNIRS and PPG, to establish a correlation model among 'behaviour, emotion, cognition, learning outcomes'.

Research Planning

Stage 1

Explore and build the basics

Build a multimodal data collection framework to conduct exploratory research on the relationship between HRV and learning states, and complete the Uedu Brain prototype and signal quality validation.

Phase 2

System development and validation

Develop the three core PALM modules (state awareness → state prediction → adaptive response), and use a quasi-experimental design to verify the intervention effect and estimate the effect size.

Stage 3

Context expansion and theory building

Cross-campus and cross-disciplinary validation, building a multimodal dynamic theoretical model of human–machine collaborative learning, and open-sourcing the reference design of Uedu Brain.

Five research questions

DescriptionRQ1: In human–AI collaborative learning processes, what dynamic patterns appear in dialogue semantics, operational behaviours and physiological signals?
CorrelationRQ2: What temporal correlations and interactions exist among the three modalities?
PredictionRQ3: Can combinations of multimodal features predict learning difficulties and learning outcomes?
InterventionRQ4: Can physiological-sensing adaptive feedback effectively reduce learning anxiety and improve learning outcomes?
TheoryRQ5: What kind of integrated information education theoretical framework is needed for human–AI collaborative learning?