Seung-Jae Kim

Seung-Jae Kim

Auto-generated profile, based on publicly available information

Seung-Jae Kim

AI-based Neurodevelopmental Diseases Digital Therapeutics Group, Korea Brain Research Institute (KBRI), 61, Cheomdan-ro, Daegu, 41062, Republic of Korea

Research Focus & Metrics

Academic Career Length

2Years (2025-Present)

(Calculated once Scilit publication data is synced: current year − year of first publication + 1)

Representative Publications

View All Publications
14 October 2025 in JMIR Serious Games
Verified
Journal Article
Open access
JKJun-Su Kim
Jun-Su Kim
YJYoo Joo Jeong
Yoo Joo Jeong
SKSeung-Jae Kim
Seung-Jae Kim
SJSu Jin Jun
Su Jin Jun
JPJin-Yeop Park
Jin-Yeop Park
JSJeong-Heon Song
Jeong-Heon Song
Background: The processing speed index (PSI) of the Korean Wechsler Intelligence Scale for Children-Fifth Edition (K-WISC-V) is highly correlated with symptoms of attention-deficit/hyperactivity disorder (ADHD) and is an important indicator of cognitive function. However, restrictions on the frequency of testing prevent short-term PSI assessments. An accessible, objective technique for predicting PSI scores would enable better short-term monitoring and intervention for children with ADHD. Objective: To enable objective and accessible monitoring of cognitive function beyond traditional clinical assessments, this study aimed to develop a machine learning model that predicts the PSI scores of children with ADHD using behavioral data from serious games. Methods: Sixty-eight children (6-13 y of age) with ADHD were recruited, and after excluding incomplete data, 59 participants were included in the final analysis. The participants completed an initial PSI assessment using the K-WISC-V followed by 25 minutes of engagement with serious game content. Data from the game sessions were used to train machine learning models, and the models’ performance in predicting PSI scores was evaluated using the root mean squared error (RMSE), mean absolute error (MAE), and mean absolute percent error (MAPE), with K-fold cross-validation (k=4) applied to ensure robustness. Results: Among the individual machine learning models, support vector regression (SVR) had the best performance, with the lowest RMSE of 11.288, MAE of 7.874, and MAPE of 7.375%. The best overall performance was achieved by the ensemble integrating AdaBoost, Elastic Net, and SVR, which recorded the lowest RMSE of 10.072, MAE of 6.798, and MAPE of 6.611%. The predictive accuracy of this ensemble model was highest for PSI scores near the mean value of 100, demonstrating its reliability for clinical applications. Conclusions: The developed PSI prediction model has the potential to serve as an objective and accessible tool for monitoring cognitive function in children with ADHD. As a complement to traditional assessments, this approach allows continuous tracking of symptom changes and can support more personalized treatment planning in both clinical and everyday settings, which may improve accessibility and adherence. However, the findings need to be validated in larger, more diverse populations, and the long-term feasibility of using serious games in clinical and educational settings must be further examined.

11 March 2025 in Clinical Pharmacology & Therapeutics
Verified
Journal Article
Open access
HLHyunju Lee
Hyunju Lee
YJYoo Joo Jeong
Yoo Joo Jeong
JKJun‐Su Kim
Jun‐Su Kim
SKSeung‐Jae Kim
Seung‐Jae Kim
UJUnhui Jo
Unhui Jo
HHHyang-Sook Hoe
Hyang-Sook Hoe
Innovations in digital technologies have emerged digital therapeutics (DTx) as a novel therapeutic intervention. DTx hold potential as novel theragnostics for disorders with broad diagnostic spectra, including neurodevelopmental diseases (NDDs). In this review, we highlight challenging factors in the successful development and deployment of DTx for NDDs with respect to patients, medical professionals, and manufacturers. We also discuss the implications of these factors and future directions for revitalizing DTx development for NDDs.

31 July 2025 in Medicinal Research Reviews
Verified
Journal Article
Open access
YJYoo Joo Jeong
Yoo Joo Jeong
HLHyunju Lee
Hyunju Lee
JKJun‐Su Kim
Jun‐Su Kim
SKSeung‐Jae Kim
Seung‐Jae Kim
UJUnhui Jo
Unhui Jo
HHHyang-Sook Hoe
Hyang-Sook Hoe
Alzheimer's disease (AD) and Parkinson's disease (PD), which are characterized by the accumulation of misfolded protein aggregates and cognitive/motor dysfunction, are the most prevalent neurodegenerative diseases (NDDs). Despite strategic investigations aimed at augmenting the pharmaceutical pipeline, available drugs for AD and PD merely slow disease progression without curing or treating the underlying pathology. Recent technological advances have given rise to digital therapeutics (DTx): software systems that aim to prevent, cure, and manage specific diseases, including NDDs. For AD and PD, the majority of DTx focus on enhancing cognitive/executive function and motor‐related functions, respectively. In this review, we describe the status of therapeutic development for AD and PD, as well as the characteristics and status of DTx targeting these NDDs. In addition, we address the limitations and challenges of DTx and their implications for the treatment of AD and PD. Ultimately, this review provides insights into the potential of DTx as a therapeutic modality and future directions for the development of DTx targeting AD and PD.

22 December 2025 in JMIR Serious Games
Verified
Journal Article
Open access
JKJun-Su Kim
Jun-Su Kim
YJYoo Joo Jeong
Yoo Joo Jeong
SKSeung-Jae Kim
Seung-Jae Kim
SJSu Jin Jun
Su Jin Jun
JPJin-Yeop Park
Jin-Yeop Park
JSJeong-Heon Song
Jeong-Heon Song
Journal of Medical Internet Research - International Scientific Journal for Medical Research, Information and Communication on the Internet #Preprint #PeerReviewMe: Warning: This is a unreviewed preprint. Readers are warned that the document has not been peer-reviewed by expert/patient reviewers or an academic editor, may contain misleading claims, and is likely to undergo changes before final publication, if accepted, or may have been rejected/withdrawn. Readers with interest and expertise are encouraged to sign up as peer-reviewer, if the paper is within an open peer-review period. Please cite this preprint only for review purposes or for grant applications and CVs (if you are the author).

11 September 2025 in JMIR Preprints
Verified
Preprint
Open access
JKJun-Su Kim
Jun-Su Kim
SKSeung-Jae Kim
Seung-Jae Kim
GRGilnam Ryu
Gilnam Ryu
SJSu-Jin Jun
Su-Jin Jun
JPJin-Yeop Park
Jin-Yeop Park
HHHyang-Sook Hoe
Hyang-Sook Hoe
BACKGROUND Recent advances in digital therapeutics (DTx) have led to the development of game-based interventions that provide engaging treatment options for individuals with attention-deficit/hyperactivity disorder (ADHD). Previous studies have mainly examined DTx usage patterns (eg, frequency and duration), however, the impact of in-game performance metrics on ADHD symptoms has not been investigated. OBJECTIVE We aimed to examine whether task difficulty levels affect ADHD symptom improvement using game-based digital therapeutics in pediatric ADHD. METHODS We analyzed data from 35 children with ADHD who participated and engaged in five cognitive digital tasks in 25-minute sessions, five times weekly for four weeks. All sessions were conducted at home using Neuro-World, an adaptive DTx platform that dynamically adjusted task difficulty based on individual performance in real time. Performance data were collected from all five cognitive tasks to derive three key indicators of task difficulty: average task difficulty (ATD), task difficulty variability (TDV), and task difficulty slope (TDS). In addition, ADHD symptom improvements were assessed using the Korean version of the ADHD Rating Scale (K-ARS) and the Korean Child Behavior Checklist (K-CBCL), administered before and after the intervention. RESULTS We found that task difficulty in game-based DTx was associated with symptom improvement in children with ADHD. Specifically, one of the five tasks, Cognitive Digital Task 1 (requiring recognition of common features among animal figures), was significantly correlated with ADHD symptom improvements. Specifically, children who performed higher levels of average task difficulty (ATD) or lower task difficulty variability (TDV) exhibited greater reductions in internalizing symptoms (K-CBCL) of ADHD. Moreover, children who progressively performed easier tasks (negative task difficulty slope [TDS]) showed more consistent improvement in hyperactivity and impulsive symptoms (K-ARS) of ADHD. CONCLUSIONS These results suggest that DTx task difficulty (ATD, TDV, TDS) differentially regulates ADHD symptoms in children with ADHD. Consequently, the consistency and quality of task performance might be critical predictors of DTx effectiveness, and task performance metrics could inform personalized intervention strategies for ADHD. CLINICALTRIAL The Korea Clinical Research Information Service KCT0009326; https://cris.nih.go.kr/cris/search/detailSearch.do?seq=26799

19 May 2026 in JMIR Formative Research
Verified
Journal Article
Open access
SLSang-Woo Lee
Sang-Woo Lee
JKJun-Su Kim
Jun-Su Kim
SKSeung-Jae Kim
Seung-Jae Kim
SCSeungho Choun
Seungho Choun
JSJeong-Heon Song
Jeong-Heon Song
Background: Cognitive decline in aging populations underscores the need for early interventions in mild cognitive impairment (MCI), where pharmacological treatments show limited benefit. Eye-movement metrics serve as sensitive markers of cognitive deficits in MCI, and digital programs integrating these tasks offer scalable, data-driven training approaches. Objective: This study aimed to evaluate the effectiveness of a digital cognitive training program incorporating eye-movement tasks in individuals with MCI, and to determine whether eye-movement indicators can serve as objective markers of cognitive improvement. Methods: A total of 12 participants aged 60-85 years with MCI (Korean version of the Montreal Cognitive Assessment [K-MoCA] score of ≤22) completed baseline and postintervention assessments using the K-MoCA and Mini-Mental State Examination-Korean version (MMSE-K). Longitudinal changes in visuospatial attention and oculomotor performance were examined using Spearman correlations across sessions, and pre-post comparisons of eye-tracking metrics were conducted to assess training-related improvements. Results: Cognitive scores improved significantly, with K-MoCA increasing by 1.5 points (from mean 20.3, SD 1.1 to mean 21.8, SD 1.7; P=.004; Cohen d=1.38) and MMSE-K by 1.3 points (from mean 21.9, SD 2.0 to mean 23.2, SD 2.2; P=.002; Cohen d=1.29). Fixation duration decreased (r=0.248; P=.003), and saccade velocity increased (r=0.258; P=.002), indicating enhanced visual processing efficiency and faster attentional shifts, whereas fixation count and saccade amplitude showed no consistent changes. In addition, saccade duration decreased by 21.72 ms, and saccade velocity increased by 114.54 °/s. Conclusions: Digital cognitive training yielded measurable gains in visuospatial attention and oculomotor efficiency in MCI, with optimized fixation and saccade patterns indicating enhanced attentional control and information processing. These findings support eye-movement metrics as sensitive indicators of cognitive change and highlight digital interventions as scalable, noninvasive tools for cognitive support in aging populations.