Minyoung Jung
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Minyoung Jung has a total of 164 co-authors. Here are the top co-authors by number of shared publications:
Yongjeon Cheong
11 shared publicationsSeonkyoung Lee
Cognitive Science Research Group, Korea Brain Research Institute, Daegu, 41068, Republic of Korea
8 shared publications- Hirotaka Kosaka
Department of Neuropsychiatry University of Fukui Fukui Japan
7 shared publications Daichi Shiotsu
Department of Neuropsychiatry, University of Fukui, Eiheiji, Japan
5 shared publicationsTaku Kamiya
Department of Neuropsychiatry, University of Fukui, Eiheiji, Japan
5 shared publicationsHidehiko Okazawa
Biomedical Imaging Research Center, University of Fukui, Eiheiji, Japan
4 shared publications
Latest Publications
- Open Access
- 8 November 2026
- Published by Elsevier in Developmental Cognitive Neuroscience
- Vol. 82
A shared somatomotor structural substrate for motor function and behavioral symptoms in childhood and adolescence
- 8 November 2026
- Published by Elsevier in Developmental Cognitive Neuroscience
- Vol. 82
Open AccessA shared somatomotor structural substrate for motor function and behavioral symptoms in childhood and adolescence
- Open Access
- 23 September 2026
- Published by JMIR Publications Inc. in JMIR Serious Games
- Vol. 14, e102714
Two-Stage Gamified Digital Assessment for Autism Spectrum Disorder Screening and the Limits of Differentiating Social Communication Disorder in Children and Adolescents: Cross-Sectional Diagnostic Accuracy Study Using Explainable Machine Learning
Background: Distinguishing autism spectrum disorder (ASD) from social communication disorder (SCD) is clinically challenging because both conditions present with overlapping social communication deficits. Standard caregiver-reported instruments capture surface-level behavioral similarities rather than underlying cognitive differences, motivating the development of digital gamified assessments that measure social cognitive processes directly. Objective: This study developed and evaluated a 2-stage gamified digital pipeline: stage 1 (Buddy Plan, a self-report module) for high-sensitivity ASD screening, and stage 2 (Buddy Drill, story-based social-judgment scenarios), which was examined with a leakage-controlled analysis, for assessing whether ASD can be differentiated from SCD. Methods: In this cross-sectional diagnostic accuracy study, 275 children and adolescents aged 6‐18 years (mean 11.07, SD 3.13 years; 175/275, 63.6% male) were recruited by convenience sampling from 5 clinical and community sites in the Republic of Korea (May 2024 to February 2025) across 5 Diagnostic and Statistical Manual of Mental Disorders, Fifth Edition (DSM-5) groups: ASD (n=51), SCD (n=54), attention-deficit/hyperactivity disorder (n=23), high risk (n=52), and neurotypically developing (ND; n=95). Diagnoses were established by board-certified child psychiatrists. Participants completed 2 tablet-based modules: Buddy Plan (52 self-report items; stage 1) and Buddy Drill (153 story-based scenarios; stage 2). The primary outcome was diagnostic accuracy (area under the receiver operating characteristic curve [AUC], sensitivity, and specificity). Four machine learning algorithms were trained with nested cross-validation (5×5 folds). For stage 2, item selection and imputation were performed within each training fold. Explainability used Shapley Additive Explanations (SHAP). Significance was set at α=.05 (2-sided) with bootstrap 95% CIs. Results: Group differences were tested by 1-way ANOVA. For stage 1 (ASD vs ND; n=146), random forest achieved a nested AUC of 0.912 (95% CI 0.856‐0.953). At a threshold of 0.200, sensitivity was 96.1% (49/51; 95% CI 86.8%‐99.5%) and specificity was 58.9% (56/95; 95% CI 48.4%‐68.9%), with 2 false negatives. For stage 2 (ASD vs SCD; n=100), the fully nested pipeline yielded only chance-level discrimination: regularized logistic regression achieved a nested AUC of 0.62 (95% CI 0.51‐0.74), and no feature configuration (self-report: 0.55, objective: 0.62, combined: 0.63) exceeded chance. SHAP identified 5 cross-algorithm stage 1 biomarkers with significant ASD-versus-ND differences (all P <.01) and no evidence of sex bias. Conclusions: The gamified Buddy Plan module shows promise for high-sensitivity ASD screening. In contrast, once feature-selection leakage was removed with a fully nested pipeline, the Buddy Drill module did not robustly differentiate ASD from SCD, and the apparent advantage of objective features over self-report features seen in leaky analyses did not persist. Because the results derive from internal cross-validation in a single, predominantly male Korean cohort without external validation or IQ matching, they represent preliminary evidence of screening feasibility rather than validated clinical differentiation. Prospective, externally validated, IQ- and language-matched studies are required.
- 23 September 2026
- Published by JMIR Publications Inc. in JMIR Serious Games
- Vol. 14, e102714
Open AccessTwo-Stage Gamified Digital Assessment for Autism Spectrum Disorder Screening and the Limits of Differentiating Social Communication Disorder in Children and Adolescents: Cross-Sectional Diagnostic Accuracy Study Using Explainable Machine Learning
Background: Distinguishing autism spectrum disorder (ASD) from social communication disorder (SCD) is clinically challenging because both conditions present with overlapping social communication deficits. Standard caregiver-reported instruments capture surface-level behavioral similarities rather than underlying cognitive differences, motivating the development of digital gamified assessments that measure social cognitive processes directly. Objective: This study developed and evaluated a 2-stage gamified digital pipeline: stage 1 (Buddy Plan, a self-report module) for high-sensitivity ASD screening, and stage 2 (Buddy Drill, story-based social-judgment scenarios), which was examined with a leakage-controlled analysis, for assessing whether ASD can be differentiated from SCD. Methods: In this cross-sectional diagnostic accuracy study, 275 children and adolescents aged 6‐18 years (mean 11.07, SD 3.13 years; 175/275, 63.6% male) were recruited by convenience sampling from 5 clinical and community sites in the Republic of Korea (May 2024 to February 2025) across 5 Diagnostic and Statistical Manual of Mental Disorders, Fifth Edition (DSM-5) groups: ASD (n=51), SCD (n=54), attention-deficit/hyperactivity disorder (n=23), high risk (n=52), and neurotypically developing (ND; n=95). Diagnoses were established by board-certified child psychiatrists. Participants completed 2 tablet-based modules: Buddy Plan (52 self-report items; stage 1) and Buddy Drill (153 story-based scenarios; stage 2). The primary outcome was diagnostic accuracy (area under the receiver operating characteristic curve [AUC], sensitivity, and specificity). Four machine learning algorithms were trained with nested cross-validation (5×5 folds). For stage 2, item selection and imputation were performed within each training fold. Explainability used Shapley Additive Explanations (SHAP). Significance was set at α=.05 (2-sided) with bootstrap 95% CIs. Results: Group differences were tested by 1-way ANOVA. For stage 1 (ASD vs ND; n=146), random forest achieved a nested AUC of 0.912 (95% CI 0.856‐0.953). At a threshold of 0.200, sensitivity was 96.1% (49/51; 95% CI 86.8%‐99.5%) and specificity was 58.9% (56/95; 95% CI 48.4%‐68.9%), with 2 false negatives. For stage 2 (ASD vs SCD; n=100), the fully nested pipeline yielded only chance-level discrimination: regularized logistic regression achieved a nested AUC of 0.62 (95% CI 0.51‐0.74), and no feature configuration (self-report: 0.55, objective: 0.62, combined: 0.63) exceeded chance. SHAP identified 5 cross-algorithm stage 1 biomarkers with significant ASD-versus-ND differences (all P <.01) and no evidence of sex bias. Conclusions: The gamified Buddy Plan module shows promise for high-sensitivity ASD screening. In contrast, once feature-selection leakage was removed with a fully nested pipeline, the Buddy Drill module did not robustly differentiate ASD from SCD, and the apparent advantage of objective features over self-report features seen in leaky analyses did not persist. Because the results derive from internal cross-validation in a single, predominantly male Korean cohort without external validation or IQ matching, they represent preliminary evidence of screening feasibility rather than validated clinical differentiation. Prospective, externally validated, IQ- and language-matched studies are required.
- Open Access
- 8 September 2026
- Published by Springer Nature in Journal of Neural Transmission
- p. 1-12
Atypical sensory traits and changes in white matter microstructures connected to the amygdala and hippocampus of the autistic brain
Autism spectrum disorder (ASD) is characterized by atypical sensory traits. The amygdala and hippocampus have been implicated in ASD, but how sensory traits relate to white matter pathways connected to these regions remains unclear. We therefore examined whether sensory traits are associated with microstructural properties of amygdala- and hippocampus-connected white matter in adults with ASD. We included 40 adults with ASD and 83 typically developing (TD) adults (all aged ≥ 18 years). Participants completed the Adolescent/Adult Sensory Profile (AASP) and underwent diffusion tensor imaging. Diffusion metrics were computed for bilateral amygdala-connected white matter (AWM) and hippocampus-connected white matter (HWM). The ASD group showed lower fractional anisotropy (FA) in both AWM and HWM than the TD group. In the ASD group, sensation-seeking scores revealed a nominal positive association with FA in the right AWM, although this association did not survive FDR correction. In the TD group, sensation-seeking scores were negatively associated with FA in the right HWM. Fisher r-to-z tests with FDR correction indicated significant group differences in the associations between sensation seeking and FA of the right AWM and right HWM. White matter microstructure in pathways connected to the amygdala and hippocampus showed group differences in FA and in how FA related to sensation seeking. These findings suggest that atypical sensory traits in ASD may be linked to microstructural differences in amygdala- and hippocampus-connected pathways.
- 8 September 2026
- Published by Springer Nature in Journal of Neural Transmission
- p. 1-12
Open AccessAtypical sensory traits and changes in white matter microstructures connected to the amygdala and hippocampus of the autistic brain
Autism spectrum disorder (ASD) is characterized by atypical sensory traits. The amygdala and hippocampus have been implicated in ASD, but how sensory traits relate to white matter pathways connected to these regions remains unclear. We therefore examined whether sensory traits are associated with microstructural properties of amygdala- and hippocampus-connected white matter in adults with ASD. We included 40 adults with ASD and 83 typically developing (TD) adults (all aged ≥ 18 years). Participants completed the Adolescent/Adult Sensory Profile (AASP) and underwent diffusion tensor imaging. Diffusion metrics were computed for bilateral amygdala-connected white matter (AWM) and hippocampus-connected white matter (HWM). The ASD group showed lower fractional anisotropy (FA) in both AWM and HWM than the TD group. In the ASD group, sensation-seeking scores revealed a nominal positive association with FA in the right AWM, although this association did not survive FDR correction. In the TD group, sensation-seeking scores were negatively associated with FA in the right HWM. Fisher r-to-z tests with FDR correction indicated significant group differences in the associations between sensation seeking and FA of the right AWM and right HWM. White matter microstructure in pathways connected to the amygdala and hippocampus showed group differences in FA and in how FA related to sensation seeking. These findings suggest that atypical sensory traits in ASD may be linked to microstructural differences in amygdala- and hippocampus-connected pathways.
- Open Access
- 7 July 2026
- Published by Springer Nature in Scientific Reports
- Vol. 16 (1)
Development and validation of the Korean Child Emotional Face Database for research in developmental neuroscience
Facial recognition plays a crucial role in identifying emotional states and understanding human responses. Research indicates that emotional processing varies based on factors such as race and age, underscoring the importance of diverse face databases. We developed the Korean Child Emotional Face Database (KCEFD), comprising facial stimuli representing seven emotions-happiness, surprise, neutral, fear, anger, disgust, and sadness-and validated it in both children and adults. The validation process involved gathering accuracy, clarity, and valence scores. Overall emotion recognition accuracy was 70%; however, while happiness (95%) and surprise (94%) showed high accuracy, fear was recognized at only 18%, which is near chance level. Happiness was rated as the most positive emotion, while anger was perceived as the most negative. Clarity scores were consistently high. To assess the effectiveness of the KCEFD as emotional stimuli, we conducted a face-in-the-crowd task using happiness, surprise, anger, and sadness expressions. By analyzing both behavioral responses and eye movements in children and adults, we confirmed the previously established anger superiority effect. The KCEFD is the first validated emotional face database specifically for Korean children, making it a valuable resource for research on children's emotional processing.
- 7 July 2026
- Published by Springer Nature in Scientific Reports
- Vol. 16 (1)
Open AccessDevelopment and validation of the Korean Child Emotional Face Database for research in developmental neuroscience
Facial recognition plays a crucial role in identifying emotional states and understanding human responses. Research indicates that emotional processing varies based on factors such as race and age, underscoring the importance of diverse face databases. We developed the Korean Child Emotional Face Database (KCEFD), comprising facial stimuli representing seven emotions-happiness, surprise, neutral, fear, anger, disgust, and sadness-and validated it in both children and adults. The validation process involved gathering accuracy, clarity, and valence scores. Overall emotion recognition accuracy was 70%; however, while happiness (95%) and surprise (94%) showed high accuracy, fear was recognized at only 18%, which is near chance level. Happiness was rated as the most positive emotion, while anger was perceived as the most negative. Clarity scores were consistently high. To assess the effectiveness of the KCEFD as emotional stimuli, we conducted a face-in-the-crowd task using happiness, surprise, anger, and sadness expressions. By analyzing both behavioral responses and eye movements in children and adults, we confirmed the previously established anger superiority effect. The KCEFD is the first validated emotional face database specifically for Korean children, making it a valuable resource for research on children's emotional processing.
- Open Access
- 26 May 2026
- Published by JMIR Publications Inc. in JMIR Serious Games
- Vol. 14, e93468
Cognitive Load Across Interaction Formats in Digital Attention Assessment for Children: Within-Subject Neuroimaging and Behavioral Comparison Study
Background: Digital health technologies increasingly use tablet-based cognitive assessments for children, yet interaction design choices can substantially influence cognitive load and measurement validity. Although cognitive load has been extensively studied in educational settings, its impact on patient-facing digital assessment tools for pediatric populations remains underexplored. Objective: This study examined how the interaction format influences cognitive load and measurement validity in a tablet-based Stroop task for children, comparing text-based with color-based response selection to determine which format better supports valid attention assessment. Methods: Using a within-subject design, 127 typically developing children (n=55, 43.3%, girls; n=72, 56.7%, boys) aged 6-12 (mean 9.15, SD 1.56) years were recruited via convenience sampling from local communities in the Republic of Korea. Participants completed both a text-based and a color-based Stroop task on a tablet. Cognitive load was indexed using prefrontal functional near-infrared spectroscopy (fNIRS), measuring functional connectivity (FC) and global network efficiency. Behavioral outcome measures included accuracy, reaction time, and the composite efficiency score. Clinical validity was assessed by correlating task performance with parent-reported attention problems using a standardized behavioral rating scale. General cognitive ability was controlled using a standardized intelligence measure. Condition differences were analyzed using paired-sample t tests (α=.05), clinical associations using Pearson correlations with Fisher r-to-z transformation, and predictor contributions using random forest regression. Results: In the paired-sample t tests, the color-based format yielded significantly higher accuracy (0.91 vs 0.86; mean difference 0.05, 95% CI 0.03-0.08; t126=3.81, Cohen d=0.34, P<.001), faster reaction times (1183 vs 1269 ms; mean difference −85.4 ms, 95% CI −107.2 to −63.5; t126=−7.72, Cohen d=–0.69, P<.001), and superior composite performance (t126=6.81, Cohen d=0.62, P<.001). Pearson correlations revealed that color-based performance was significantly associated with parent-reported attention problems (r=−0.20, 95% CI −0.37 to −0.03; P=.03), whereas text-based performance was not (r=−0.05, 95% CI −0.23 to 0.12;’ P=.56); the Fisher r-to-z test indicated this difference was not statistically significant (z=1.54, P=.12). fNIRS-derived neural indices (ΔFC, global efficiency) did not differ significantly between conditions (P>.05). Random forest analyses indicated that after controlling for age and general cognitive ability, individual variations in prefrontal efficiency accounted for 40% to 44% of residual performance variance. Conclusions: This study provides some of the first empirical evidence that the interaction format substantially influences task demands and measurement validity in pediatric digital assessments, extending cognitive load theory into digital health assessment design. Color-based response formats that minimize extraneous semantic processing yield superior performance and stronger clinical validity compared to text-based formats. These findings suggest that prioritizing response modalities aligned with children’s developmental capabilities may improve the clinical utility of digital attention assessments.
- 26 May 2026
- Published by JMIR Publications Inc. in JMIR Serious Games
- Vol. 14, e93468
Open AccessCognitive Load Across Interaction Formats in Digital Attention Assessment for Children: Within-Subject Neuroimaging and Behavioral Comparison Study
Background: Digital health technologies increasingly use tablet-based cognitive assessments for children, yet interaction design choices can substantially influence cognitive load and measurement validity. Although cognitive load has been extensively studied in educational settings, its impact on patient-facing digital assessment tools for pediatric populations remains underexplored. Objective: This study examined how the interaction format influences cognitive load and measurement validity in a tablet-based Stroop task for children, comparing text-based with color-based response selection to determine which format better supports valid attention assessment. Methods: Using a within-subject design, 127 typically developing children (n=55, 43.3%, girls; n=72, 56.7%, boys) aged 6-12 (mean 9.15, SD 1.56) years were recruited via convenience sampling from local communities in the Republic of Korea. Participants completed both a text-based and a color-based Stroop task on a tablet. Cognitive load was indexed using prefrontal functional near-infrared spectroscopy (fNIRS), measuring functional connectivity (FC) and global network efficiency. Behavioral outcome measures included accuracy, reaction time, and the composite efficiency score. Clinical validity was assessed by correlating task performance with parent-reported attention problems using a standardized behavioral rating scale. General cognitive ability was controlled using a standardized intelligence measure. Condition differences were analyzed using paired-sample t tests (α=.05), clinical associations using Pearson correlations with Fisher r-to-z transformation, and predictor contributions using random forest regression. Results: In the paired-sample t tests, the color-based format yielded significantly higher accuracy (0.91 vs 0.86; mean difference 0.05, 95% CI 0.03-0.08; t126=3.81, Cohen d=0.34, P<.001), faster reaction times (1183 vs 1269 ms; mean difference −85.4 ms, 95% CI −107.2 to −63.5; t126=−7.72, Cohen d=–0.69, P<.001), and superior composite performance (t126=6.81, Cohen d=0.62, P<.001). Pearson correlations revealed that color-based performance was significantly associated with parent-reported attention problems (r=−0.20, 95% CI −0.37 to −0.03; P=.03), whereas text-based performance was not (r=−0.05, 95% CI −0.23 to 0.12;’ P=.56); the Fisher r-to-z test indicated this difference was not statistically significant (z=1.54, P=.12). fNIRS-derived neural indices (ΔFC, global efficiency) did not differ significantly between conditions (P>.05). Random forest analyses indicated that after controlling for age and general cognitive ability, individual variations in prefrontal efficiency accounted for 40% to 44% of residual performance variance. Conclusions: This study provides some of the first empirical evidence that the interaction format substantially influences task demands and measurement validity in pediatric digital assessments, extending cognitive load theory into digital health assessment design. Color-based response formats that minimize extraneous semantic processing yield superior performance and stronger clinical validity compared to text-based formats. These findings suggest that prioritizing response modalities aligned with children’s developmental capabilities may improve the clinical utility of digital attention assessments.
- Open Access
- 25 May 2026
- Published by Elsevier in Journal of Affective Disorders
- Vol. 413, 122001
Stressed-out parenting brain: Effects of parenting stress, depression and anxiety on caregivers' brain structure and function
- 25 May 2026
- Published by Elsevier in Journal of Affective Disorders
- Vol. 413, 122001
Open AccessStressed-out parenting brain: Effects of parenting stress, depression and anxiety on caregivers' brain structure and function
- Open Access
- 22 February 2026
- Published by JMIR Publications Inc. in JMIR Preprints
Cognitive Load Across Interaction Formats in Digital Attention Assessment for Children: Within-Subject Neuroimaging and Behavioral Comparison Study (Preprint)
BACKGROUND Digital health technologies increasingly use tablet-based cognitive assessments for children, yet interaction design choices can substantially influence cognitive load and measurement validity. Although cognitive load has been extensively studied in educational settings, its impact on patient-facing digital assessment tools for pediatric populations remains underexplored. OBJECTIVE This study examined how the interaction format influences cognitive load and measurement validity in a tablet-based Stroop task for children, comparing text-based with color-based response selection to determine which format better supports valid attention assessment. METHODS Using a within-subject design, 127 typically developing children (n=55, 43.3%, girls; n=72, 56.7%, boys) aged 6-12 (mean 9.15, SD 1.56) years were recruited via convenience sampling from local communities in the Republic of Korea. Participants completed both a text-based and a color-based Stroop task on a tablet. Cognitive load was indexed using prefrontal functional near-infrared spectroscopy (fNIRS), measuring functional connectivity (FC) and global network efficiency. Behavioral outcome measures included accuracy, reaction time, and the composite efficiency score. Clinical validity was assessed by correlating task performance with parent-reported attention problems using a standardized behavioral rating scale. General cognitive ability was controlled using a standardized intelligence measure. Condition differences were analyzed using paired-sample t tests (α=.05), clinical associations using Pearson correlations with Fisher r-to-z transformation, and predictor contributions using random forest regression. RESULTS In the paired-sample t tests, the color-based format yielded significantly higher accuracy (0.91 vs 0.86; mean difference 0.05, 95% CI 0.03-0.08; t126=3.81, Cohen d=0.34, P<.001), faster reaction times (1183 vs 1269 ms; mean difference −85.4 ms, 95% CI −107.2 to −63.5; t126=−7.72, Cohen d=–0.69, P<.001), and superior composite performance (t126=6.81, Cohen d=0.62, P<.001). Pearson correlations revealed that color-based performance was significantly associated with parent-reported attention problems (r=−0.20, 95% CI −0.37 to −0.03; P=.03), whereas text-based performance was not (r=−0.05, 95% CI −0.23 to 0.12;’ P=.56); the Fisher r-to-z test indicated this difference was not statistically significant (z=1.54, P=.12). fNIRS-derived neural indices (ΔFC, global efficiency) did not differ significantly between conditions (P>.05). Random forest analyses indicated that after controlling for age and general cognitive ability, individual variations in prefrontal efficiency accounted for 40% to 44% of residual performance variance. CONCLUSIONS This study provides some of the first empirical evidence that the interaction format substantially influences task demands and measurement validity in pediatric digital assessments, extending cognitive load theory into digital health assessment design. Color-based response formats that minimize extraneous semantic processing yield superior performance and stronger clinical validity compared to text-based formats. These findings suggest that prioritizing response modalities aligned with children’s developmental capabilities may improve the clinical utility of digital attention assessments. CLINICALTRIAL
- 22 February 2026
- Published by JMIR Publications Inc. in JMIR Preprints
Open AccessCognitive Load Across Interaction Formats in Digital Attention Assessment for Children: Within-Subject Neuroimaging and Behavioral Comparison Study (Preprint)
BACKGROUND Digital health technologies increasingly use tablet-based cognitive assessments for children, yet interaction design choices can substantially influence cognitive load and measurement validity. Although cognitive load has been extensively studied in educational settings, its impact on patient-facing digital assessment tools for pediatric populations remains underexplored. OBJECTIVE This study examined how the interaction format influences cognitive load and measurement validity in a tablet-based Stroop task for children, comparing text-based with color-based response selection to determine which format better supports valid attention assessment. METHODS Using a within-subject design, 127 typically developing children (n=55, 43.3%, girls; n=72, 56.7%, boys) aged 6-12 (mean 9.15, SD 1.56) years were recruited via convenience sampling from local communities in the Republic of Korea. Participants completed both a text-based and a color-based Stroop task on a tablet. Cognitive load was indexed using prefrontal functional near-infrared spectroscopy (fNIRS), measuring functional connectivity (FC) and global network efficiency. Behavioral outcome measures included accuracy, reaction time, and the composite efficiency score. Clinical validity was assessed by correlating task performance with parent-reported attention problems using a standardized behavioral rating scale. General cognitive ability was controlled using a standardized intelligence measure. Condition differences were analyzed using paired-sample t tests (α=.05), clinical associations using Pearson correlations with Fisher r-to-z transformation, and predictor contributions using random forest regression. RESULTS In the paired-sample t tests, the color-based format yielded significantly higher accuracy (0.91 vs 0.86; mean difference 0.05, 95% CI 0.03-0.08; t126=3.81, Cohen d=0.34, P<.001), faster reaction times (1183 vs 1269 ms; mean difference −85.4 ms, 95% CI −107.2 to −63.5; t126=−7.72, Cohen d=–0.69, P<.001), and superior composite performance (t126=6.81, Cohen d=0.62, P<.001). Pearson correlations revealed that color-based performance was significantly associated with parent-reported attention problems (r=−0.20, 95% CI −0.37 to −0.03; P=.03), whereas text-based performance was not (r=−0.05, 95% CI −0.23 to 0.12;’ P=.56); the Fisher r-to-z test indicated this difference was not statistically significant (z=1.54, P=.12). fNIRS-derived neural indices (ΔFC, global efficiency) did not differ significantly between conditions (P>.05). Random forest analyses indicated that after controlling for age and general cognitive ability, individual variations in prefrontal efficiency accounted for 40% to 44% of residual performance variance. CONCLUSIONS This study provides some of the first empirical evidence that the interaction format substantially influences task demands and measurement validity in pediatric digital assessments, extending cognitive load theory into digital health assessment design. Color-based response formats that minimize extraneous semantic processing yield superior performance and stronger clinical validity compared to text-based formats. These findings suggest that prioritizing response modalities aligned with children’s developmental capabilities may improve the clinical utility of digital attention assessments. CLINICALTRIAL
- Open Access
A Training-Free Large Reasoning Model-based Knowledge Tracing Framework for Unified Prediction and Prescription
Knowledge Tracing (KT) aims to estimate a learner's evolving mastery based on interaction histories. Recent studies have explored Large Language Models (LLMs) for KT via autoregressive nature, but such approaches typically require fine-tuning and exhibit unstable or near-random performance. Moreover, prior KT systems primarily focus on prediction and rely on multi-stage pipelines for feedback and recommendation, resulting in increased system complexity and resources. To address this gap, we propose Thinking-KT, a training-free KT framework that incorporates Test-Time Scaling (TTS), enabling even small LLMs to achieve competitive KT performance. Moreover, in this framework, a small LLM can jointly perform KT prediction, personalized feedback generation, and learning recommendation in a unified output without degrading prediction accuracy. Beyond performance, we present the systematic analysis of reasoning traces in KT. Our results demonstrate that TTS is a critical yet underexplored factor in LLM-based KT, and that small LLMs can serve as unified ITS engines.
Open AccessA Training-Free Large Reasoning Model-based Knowledge Tracing Framework for Unified Prediction and Prescription
Knowledge Tracing (KT) aims to estimate a learner's evolving mastery based on interaction histories. Recent studies have explored Large Language Models (LLMs) for KT via autoregressive nature, but such approaches typically require fine-tuning and exhibit unstable or near-random performance. Moreover, prior KT systems primarily focus on prediction and rely on multi-stage pipelines for feedback and recommendation, resulting in increased system complexity and resources. To address this gap, we propose Thinking-KT, a training-free KT framework that incorporates Test-Time Scaling (TTS), enabling even small LLMs to achieve competitive KT performance. Moreover, in this framework, a small LLM can jointly perform KT prediction, personalized feedback generation, and learning recommendation in a unified output without degrading prediction accuracy. Beyond performance, we present the systematic analysis of reasoning traces in KT. Our results demonstrate that TTS is a critical yet underexplored factor in LLM-based KT, and that small LLMs can serve as unified ITS engines.
- Open Access
- 1 September 2025
- Published by Springer Nature in Translational Psychiatry
- Vol. 15 (1) , 1-11
Dissecting the heterogeneity of autism spectrum disorder with sensory behavior, brain, and epigenetic factors
Autism spectrum disorder (ASD), a disorder with heterogeneous etiology, is characterized by abnormal behavioral responses to sensory inputs. However, there is still limited understanding of how brain and epigenetic factors, along with behavioral abnormality, contribute to ASD. After completing Adolescent-Adult Sensory Profile, a self-report questinnaire, 34 individuals with ASD and 72 controls underwent neuroimaging scans to measure brain structural (cortical and subcortical volume) and functional (thalamo-cortical resting-state functional connectivity) characteristics. For epigenetic measures, we computed DNA methylation values of the oxytocin receptor and arginine vasopressin receptor (AVPR) genes from the participants’ saliva. When sensory-related behavior was the default baseline, a machine learning algorithm demonstrated that the neuroimaging-epigenetic model outperformed the neuroimaging model or the epigenetic model. Thalamo-cortical hyperconnectivity and AVPR 1A epigenetic modification were found to be significant contributing factors in these models. By integrating neuroimaging and epigenetic biomarkers with behaviors, a more precise diagnosis of ASD can be achieved.
- 1 September 2025
- Published by Springer Nature in Translational Psychiatry
- Vol. 15 (1) , 1-11
Open AccessDissecting the heterogeneity of autism spectrum disorder with sensory behavior, brain, and epigenetic factors
Autism spectrum disorder (ASD), a disorder with heterogeneous etiology, is characterized by abnormal behavioral responses to sensory inputs. However, there is still limited understanding of how brain and epigenetic factors, along with behavioral abnormality, contribute to ASD. After completing Adolescent-Adult Sensory Profile, a self-report questinnaire, 34 individuals with ASD and 72 controls underwent neuroimaging scans to measure brain structural (cortical and subcortical volume) and functional (thalamo-cortical resting-state functional connectivity) characteristics. For epigenetic measures, we computed DNA methylation values of the oxytocin receptor and arginine vasopressin receptor (AVPR) genes from the participants’ saliva. When sensory-related behavior was the default baseline, a machine learning algorithm demonstrated that the neuroimaging-epigenetic model outperformed the neuroimaging model or the epigenetic model. Thalamo-cortical hyperconnectivity and AVPR 1A epigenetic modification were found to be significant contributing factors in these models. By integrating neuroimaging and epigenetic biomarkers with behaviors, a more precise diagnosis of ASD can be achieved.
- Open Access
Deformation-Aware Segmentation Network Robust to Motion Artifacts for Brain Tissue Segmentation using Disentanglement Learning
Motion artifacts caused by prolonged acquisition time are a significant challenge in Magnetic Resonance Imaging (MRI), hindering accurate tissue segmentation. These artifacts appear as blurred images that mimic tissue-like appearances, making segmentation difficult. This study proposes a novel deep learning framework that demonstrates superior performance in both motion correction and robust brain tissue segmentation in the presence of artifacts. The core concept lies in a complementary process: a disentanglement learning network progressively removes artifacts, leading to cleaner images and consequently, more accurate segmentation by a jointly trained motion estimation and segmentation network. This network generates three outputs: a motioncorrected image, a motion deformation map that identifies artifact-affected regions, and a brain tissue segmentation mask. This deformation serves as a guidance mechanism for the disentanglement process, aiding the model in recovering lost information or removing artificial structures introduced by the artifacts. Extensive in-vivo experiments on pediatric motion data demonstrate that our proposed framework outperforms state-of-the-art methods in segmenting motion-corrupted MRI scans.
Open AccessDeformation-Aware Segmentation Network Robust to Motion Artifacts for Brain Tissue Segmentation using Disentanglement Learning
Motion artifacts caused by prolonged acquisition time are a significant challenge in Magnetic Resonance Imaging (MRI), hindering accurate tissue segmentation. These artifacts appear as blurred images that mimic tissue-like appearances, making segmentation difficult. This study proposes a novel deep learning framework that demonstrates superior performance in both motion correction and robust brain tissue segmentation in the presence of artifacts. The core concept lies in a complementary process: a disentanglement learning network progressively removes artifacts, leading to cleaner images and consequently, more accurate segmentation by a jointly trained motion estimation and segmentation network. This network generates three outputs: a motioncorrected image, a motion deformation map that identifies artifact-affected regions, and a brain tissue segmentation mask. This deformation serves as a guidance mechanism for the disentanglement process, aiding the model in recovering lost information or removing artificial structures introduced by the artifacts. Extensive in-vivo experiments on pediatric motion data demonstrate that our proposed framework outperforms state-of-the-art methods in segmenting motion-corrupted MRI scans.