Hanyang University Professor Noh Youngtae's Team Develops iEMA, Selected for Samsung Project

Key Achievements in Next-Generation Electronic Device Core Technologies

Hanyang University announced on October 24 that the research project "iEMA: Intelligent Optimal Survey Timing Detection-Based Psychological and Behavioral Data Collection Platform," proposed by the research team of Professor Noh Youngtae from the Department of Data Science and the Department of Artificial Intelligence (including integrated master's and doctoral program student Lee Hyesung, doctoral student Jin Seungwan, and integrated master's and doctoral program student Alfred), has been selected as a finalist in the designated theme competition for the 2025 Samsung Future Technology Development Program.

(Top row from left) Jin Seungwan, Alfred (Bottom row from left) Professor Noh Youngtae, Lee Hyesung. Hanyang University

The Samsung Future Technology Development Program is South Korea's leading private research support initiative, which has been funding "high-risk, high-impact" research in the fields of basic science, materials, and ICT since 2013. To date, the program has invested more than 1 trillion won in approximately 500 projects, contributing to the discovery of world-class fundamental technologies.

In particular, the designated theme competition is a top-tier competitive program that evaluates both social impact and academic innovation. The selection of Professor Noh Youngtae's team officially recognizes the originality and technological potential of their research.

The team's project, "iEMA: Intelligent Optimal Survey Timing Detection-Based Psychological and Behavioral Data Collection Platform," aims to collect physiological and behavioral data from daily life using smartphones and wearable devices, and to build a new type of intelligent data labeling framework (iEMA) in which artificial intelligence autonomously determines the optimal timing and necessity for surveys.

Traditional Ecological Momentary Assessment (EMA) collects daily data through periodic surveys, but it has limitations such as the mismatch between symptom onset and survey timing, accumulated fatigue from repeated surveys, and reduced label reliability.

To address these issues, the research team developed two core modules: "Significant Opportune Moment (SOM)" and "Survey Fatigue Minimization (SFM)." By detecting the optimal moment immediately after symptom onset when the user is available to respond, they plan to dramatically improve the reliability of data labeling.

The core of iEMA lies in the real-time analysis of physiological and behavioral data collected from smartphones and wearable devices by artificial intelligence, which autonomously determines whether "now is the optimal moment to present a survey." To achieve this, the team developed the "SOM (Significant Opportune Moment)" module, which detects the appropriate timing for surveys immediately after symptom onset, and the "SFM (Survey Fatigue Minimization)" module, which minimizes survey fatigue. By combining these two modules, unnecessary surveys can be reduced while securing high-quality data that aligns with the actual timing of symptom occurrence.

Additionally, the team applied active learning (ALS) and incremental learning (EIL) technologies to build an algorithmic structure that reduces the number of surveys by 60% to 90% while maintaining data quality. By gradually learning users' physiological and behavioral patterns over time, the system is designed to evolve into a personalized stress detection model.

The iEMA project will be conducted over three years, from December 2025 to November 2028, with a total budget of 1.2 billion won. The project aims to collect more than 8,000 weeks of data from approximately 600 participants. Through an app based on smartwatches and smartphones, it will integrate the collection of heart rate, activity, sleep, and environmental data, which will then be analyzed in real time using the university's cloud platform, "EasyTrack."

This research is expected to drive innovation in data labeling necessary for the development of digital therapeutics (DTx) based on daily life. It also has significant potential for expansion into various fields, including mental health management, early stress detection, and long-term well-being data analysis.

Professor Noh Youngtae stated, "This research is the first attempt to design artificial intelligence that can more accurately understand a person's condition and autonomously determine 'when to ask' questions." He added, "In the future, we aim to expand this into an on-device psychological and behavioral data collection platform, contributing to the establishment of a national mental health big data infrastructure."

This content was produced with the assistance of AI translation services.