Machine Learning-Based Prediction of Cognitive Function in Alzheimer’s Disease Using Multimodal Data
Poster No:
175
Submission Type:
Abstract Submission
Authors:
Choonghee Park1, Kyeong Seob Song1, Hyun Kook Lim2,3, Dae-Jin Kim4, Tae-Min Kim1,3, Ji-Won Chun1,3
Institutions:
1Department of Medical Informatics, College of Medicine, The Catholic University of Korea, Seoul, Republic of Korea, 2Department of Psychiatry, Yeouido St. Mary's Hospital, Seoul, Republic of Korea, 3CMC Institute for Basic Medical Science, The Catholic Medical Center of The Catholic University of Korea, Seoul, Republic of Korea, 4Department of Psychiatry, Seoul St. Mary's Hospital, Seoul, Republic of Korea
First Author:
Choonghee Park
Department of Medical Informatics, College of Medicine, The Catholic University of Korea
Seoul, Republic of Korea
Department of Medical Informatics, College of Medicine, The Catholic University of Korea
Seoul, Republic of Korea
Co-Author(s):
Kyeongseob Song
Department of Medical Informatics, College of Medicine, The Catholic University of Korea
Seoul, Republic of Korea
Department of Medical Informatics, College of Medicine, The Catholic University of Korea
Seoul, Republic of Korea
Hyun Kook Lim
Department of Psychiatry, Yeouido St. Mary's Hospital|CMC Institute for Basic Medical Science, The Catholic Medical Center of The Catholic University of Korea
Seoul, Republic of Korea|Seoul, Republic of Korea
Department of Psychiatry, Yeouido St. Mary's Hospital|CMC Institute for Basic Medical Science, The Catholic Medical Center of The Catholic University of Korea
Seoul, Republic of Korea|Seoul, Republic of Korea
Tae-Min Kim
Department of Medical Informatics, College of Medicine, The Catholic University of Korea|CMC Institute for Basic Medical Science, The Catholic Medical Center of The Catholic University of Korea
Seoul, Republic of Korea|Seoul, Republic of Korea
Department of Medical Informatics, College of Medicine, The Catholic University of Korea|CMC Institute for Basic Medical Science, The Catholic Medical Center of The Catholic University of Korea
Seoul, Republic of Korea|Seoul, Republic of Korea
Ji-Won Chun
Department of Medical Informatics, College of Medicine, The Catholic University of Korea|CMC Institute for Basic Medical Science, The Catholic Medical Center of The Catholic University of Korea
Seoul, Republic of Korea|Seoul, Republic of Korea
Department of Medical Informatics, College of Medicine, The Catholic University of Korea|CMC Institute for Basic Medical Science, The Catholic Medical Center of The Catholic University of Korea
Seoul, Republic of Korea|Seoul, Republic of Korea
Introduction:
Cognitive impairment in Alzheimer's disease (AD) progresses unevenly, with memory often declining first and other domains, such as language and executive function, affected later. Understanding these domain-specific trajectories can guide targeted interventions and personalized management. This study proposes regression models integrating MRI-derived features and clinical data to predict standardized neuropsychological scores for individual cognitive domains. Previous work shows that machine learning models combining MRI, cognitive assessments, and clinical biomarkers can significantly improve predictions of cognitive decline. By employing these multimodal approaches, this analysis seeks to deepen understanding of how specific cognitive domains evolve in relation to underlying neuroanatomical changes. Ultimately, this may inform more effective clinical decision-making and earlier, more personalized interventions in AD.
Methods:
This study utilized data from 4,092 participants in the AI Hub platform provided by the National Information Society Agency of Korea (NIA), comprising cognitively normal individuals as well as those with mild cognitive impairment (MCI) and Alzheimer's disease (AD). Multimodal inputs included T1-weighted MRI scans, clinical information, and dementia diagnosis labels. MRI preprocessing involved FreeSurfer segmentation into 104 brain regions according to the Desikan-Killiany atlas, yielding volumetric, thickness, and curvature metrics. Additional features were engineered, such as total brain and cortex volumes, hippocampal proportions, and hippocampus-to-ventricle ratios, to capture nuanced atrophy patterns associated with AD. Standardized z-scores from neuropsychological assessments (SNSB, CERAD-K) provided comparable outcome measures across cognitive domains (Language, Visuospatial, Memory, Frontal/Executive). A multi-output regression framework employed LightGBM, SVR, and Random Forest as base learners, and a Linear Regression meta-model for stacked predictions. Model selection was guided by five-fold cross-validation.
·Comparison of performances of multimodal data
Results:
The stacking ensemble, incorporating LGBM, SVR, and Random Forest as base learners with Linear Regression as the meta-model, demonstrated improved predictive performance when integrating multimodal data. Across all cognitive domains, combining clinical data with MRI features (bimodal approach) resulted in higher predictive accuracy compared to using MRI data alone (unimodal approach) across different cognitive domains: Language (R2 = 0.48; R2 = 0.37, respectively), Visuospatial (R2 = 0.40; R2 = 0.21, respectively), Memory (R2 = 0.52; R2 = 0.35, respectively), and Frontal/Executive (R2 = 0.58; R2 = 0.37, respectively). This enhancement in predictive accuracy suggests that multimodal input, combining neuroanatomical measures with clinically relevant dementia indicators, contributes to more robust estimates of cognitive function in individuals with Alzheimer's disease. In addition, distinct anatomical patterns emerged for each cognitive domain. For instance, language prediction was strongly associated with age, education, specific subcortical structures (e.g., thalamus, putamen, amygdala, hippocampus), and global measures like MMSE. Similarly, visuospatial and memory performance were linked to both cortical and subcortical brain measures, as well as education and clinical scores, while frontal/executive functioning emphasized connections with various corpus callosum regions and hippocampal ratios.
·A flow diagram illustrating data selection criteria from AI Hub
Conclusions:
These findings not only advance the understanding of how specific cognitive domains evolve over time in relation to underlying brain changes, but also suggest that tailored, domain-specific intervention strategies may be more effective in clinical practice. Ultimately, this multimodal, machine learning-based approach offers a more nuanced and actionable framework for predicting cognitive impairment, informing earlier and more personalized management strategies in Alzheimer's disease.
Disorders of the Nervous System:
Neurodegenerative/ Late Life (eg. Parkinson’s, Alzheimer’s) 1
Modeling and Analysis Methods:
Classification and Predictive Modeling 2
Multivariate Approaches
Neuroinformatics and Data Sharing:
Databasing and Data Sharing
Novel Imaging Acquisition Methods:
Anatomical MRI
Keywords:
Degenerative Disease
Machine Learning
Modeling
STRUCTURAL MRI
Other - Cognitive Function
1|2Indicates the priority used for review
Abstract Information
By submitting your proposal, you grant permission for the Organization for Human Brain Mapping (OHBM) to distribute your work in any format, including video, audio print and electronic text through OHBM OnDemand, social media channels, the OHBM website, or other electronic publications and media.
The Open Science Special Interest Group (OSSIG) is introducing a reproducibility challenge for OHBM 2025. This new initiative aims to enhance the reproducibility of scientific results and foster collaborations between labs. Teams will consist of a “source” party and a “reproducing” party, and will be evaluated on the success of their replication, the openness of the source work, and additional deliverables. Click here for more information. Propose your OHBM abstract(s) as source work for future OHBM meetings by selecting one of the following options:
Please indicate below if your study was a "resting state" or "task-activation” study.
Healthy subjects only or patients (note that patient studies may also involve healthy subjects):
Was this research conducted in the United States?
Were any human subjects research approved by the relevant Institutional Review Board or ethics panel? NOTE: Any human subjects studies without IRB approval will be automatically rejected.
Were any animal research approved by the relevant IACUC or other animal research panel? NOTE: Any animal studies without IACUC approval will be automatically rejected.
Please indicate which methods were used in your research:
For human MRI, what field strength scanner do you use?
Which processing packages did you use for your study?
Provide references using APA citation style.
[2] Gorji, A., & Fathi Jouzdani, A. (2024). Machine learning for predicting cognitive decline within five years in Parkinson’s disease: Comparing cognitive assessment scales with DAT SPECT and clinical biomarkers. Plos one, 19(7), e0304355.
[3] Ryu, H. J., & Yang, D. W. (2023). The Seoul neuropsychological screening battery (SNSB) for comprehensive neuropsychological assessment. Dementia and Neurocognitive Disorders, 22(1), 1.
[4] Moon, H., Lee, E. S., Na, S., An, D., Shin, J. S., Na, D. L., & Jang, H. (2024). Discriminative Power of Seoul Cognitive Status Test in Differentiating Subjective Cognitive Decline, Amnestic Mild Cognitive Impairment, and Dementia Based on CERAD-K Standards. Dementia and Neurocognitive Disorders, 23(3), 136.
[5] Harasty, J. A., Halliday, G. M., Kril, J. J., & Code, C. (1999). Specific temporoparietal gyral atrophy reflects the pattern of language dissolution in Alzheimer's disease. Brain, 122(4), 675-686.
[6] Szatloczki, G., Hoffmann, I., Vincze, V., Kalman, J., & Pakaski, M. (2015). Speaking in Alzheimer’s disease, is that an early sign? Importance of changes in language abilities in Alzheimer’s disease. Frontiers in aging neuroscience, 7, 195.
[7] Possin, K. L. (2010). Visual spatial cognition in neurodegenerative disease. Neurocase, 16(6), 466-487.
[8] Neufang, S., Akhrif, A., Riedl, V., Förstl, H., Kurz, A., Zimmer, C., ... & Wohlschläger, A. M. (2011). Disconnection of frontal and parietal areas contributes to impaired attention in very early Alzheimer's disease. Journal of Alzheimer's Disease, 25(2), 309-321.
[9] Basso, M., Yang, J., Warren, L., MacAvoy, M. G., Varma, P., Bronen, R. A., & van Dyck, C. H. (2006). Volumetry of amygdala and hippocampus and memory performance in Alzheimer's disease. Psychiatry Research: Neuroimaging, 146(3), 251-261.
[10] Frisoni, G. B., Fox, N. C., Jack Jr, C. R., Scheltens, P., & Thompson, P. M. (2010). The clinical use of structural MRI in Alzheimer disease. Nature reviews neurology, 6(2), 67-77.
[11] Bartos, A., Gregus, D., Ibrahim, I., & Tintěra, J. (2019). Brain volumes and their ratios in Alzheimer s disease on magnetic resonance imaging segmented using Freesurfer 6.0. Psychiatry Research: Neuroimaging, 287, 70-74.