First, we established an experimental protocol for quantitatively evaluating psychological and physiological responses when sensitive personal data are presented. In this study, we aim to clarify how psychological shock, anxiety, positive acceptance, and other reactions occur when sensitive information such as “epigenomic age” is presented to the person concerned. For this purpose, as shown in the upper right figure, whereas conventionally there was no choice but to convey information directly to the patient, we predict in advance the response when the information is conveyed through the process of grasping characteristics and measuring physiological responses. For grasping characteristics, we created questionnaire items including “degree of trust in science” and other items.

From this, we began exploring factors that explain individual differences, such as people who are likely to show anxiety in response to information presentation and people who are likely to accept it positively. In addition, we considered that the response when information is actually conveyed could be predicted by measuring physiological responses while viewing similar explanatory videos and other materials. In an initial experiment, it was suggested that electrodermal activity responses during viewing of a concept explanation video may be able to predict subsequent responses at the time of result presentation. This outcome becomes a basis for estimating in advance the psychosomatic effects of sensitive information presentation and for adjusting the presented content, expression, and timing.
Second, as a technology for generating personalized AI intervention content using generative AI, we developed a positive movie generation method. This approach generates videos of “oneself taking desirable actions” or “oneself smoothly conversing with an unknown person” using the individual’s face, body, and voice. By viewing these videos in advance of real situations, the aim is to reduce tension and anxiety while enhancing motivation for action. In particular, we designed an experimental protocol to test whether, prior to meeting an unfamiliar counselor, viewing a video of “oneself smoothly conversing with that counselor” can reduce anxiety and tension during the subsequent real interaction. This technology is expected to function as an intervention method that lowers psychological barriers before individuals engage in support settings such as counseling, education, welfare, medical care, and social skills training.
Third, we developed presentation technology for self-images and images of others in online and AR environments. In online conversation, because nonverbal information such as gaze, facial expressions, and face orientation is more limited than in face-to-face conversation, the impression given to the other party and the smoothness of the conversation may decline. Therefore, we created an automatic mirroring technology that recognizes the facial movements of the conversation partner in real time and automatically synthesizes corresponding movements onto the user’s facial video (figure below). In addition, in order to reduce the impression gap between an edited face online and the real face in face-to-face situations, we developed a system that gradually transitions from the edited face to the real face using AR. These results are foundational technologies for realizing a more natural and easily acceptable communication environment by suppressing interpersonal anxiety and discomfort in online consultation and remote support.