Progress Report

Construction of an AIoT-based universal emotional state space and evaluation of well-/ill-being states[3] Establishment of Well-being/Iii-being Detection Technology

Progress until FY2024

1. Outline of the project

Based on the outcomes of the research and development themes "Constructing a Human Emotional State Space by AIoT" and "Constructing a Universal Emotional State Space," we will develop technology to detect and evaluate mental well-being by characterizing the dynamical properties of emotional state transitions, such as transition matrices, transition frequencies, residence times, or first passage times) (Figure 1).

Fig.1
Figure 1

2. Outcome so far

[Main Achievements by FY2023]

As a support technology for mental well-being, we analyzed research data obtained from an intervention study that monitored and evaluated daily sleep using the sleep tracker-linked cloud system developed under the research and development theme "Constructing a Human Emotional State Space by AIoT." Random push notifications aimed at improving sleep were sent to groups with unstable and stable sleep patterns (Figure 2). This intervention study is positioned as a Just-In-Time Adaptive Intervention using micro-randomization. As a result, feedback on the previous night's sleep duration led to a significant extension of sleep duration and improvements in mood and sleepiness upon waking. Furthermore, these effects were observed only in the unstable group.

Fig.
Figure 2
[Main Achievements in FY2024]

We identified factors associated with daily presenteeism and psycho-physiological conditions, publishing these findings as an academic paper (Figure 2). These insights contribute to specifying intervention targets aimed at enhancing workplace productivity. Additionally, we developed technology using an IoT cloud system to detect sleep vulnerability among workers and conducted personalized interventions utilizing mobile technology, demonstrating for the first time globally a significant improvement in sleep quality (Figure 3), with results published academically. Furthermore, we performed a two-month field study involving 158 workers, collecting and analyzing approximately 34,800 questionnaire responses and wearable data points (~8,400 person-days), and successfully identified indicators predictive of transitions to ill-being states.

Fig.
Figure 3

3. Future plans

We aim to establish technology for the early detection and monitoring of mental well-being and ill-being states by analyzing the dynamic characteristics of emotional state transitions in healthy individuals, patients with mental disorders, and disease-model mice.

(NAKAMURA Toru: The University of Osaka
YAMAMOTO Yoshiharu: The University of Tokyo)