{"status":"ok","message-type":"work","message-version":"1.0.0","message":{"institution":[{"name":"medRxiv"}],"indexed":{"date-parts":[[2026,8,29]],"date-time":"2026-08-29T10:39:07Z","timestamp":1787999947205,"version":"build-2784847793"},"posted":{"date-parts":[[2026,4,15]]},"group-title":"Health Informatics","reference-count":49,"publisher":"openRxiv","license":[{"start":{"date-parts":[[2026,4,15]],"date-time":"2026-04-15T00:00:00Z","timestamp":1776211200000},"content-version":"vor","delay-in-days":0,"URL":"https:\/\/www.medrxiv.org\/about\/FAQ#license"}],"content-domain":{"domain":[],"crossmark-restriction":false},"short-container-title":[],"accepted":{"date-parts":[[2026,4,15]]},"abstract":"<jats:title>Abstract<\/jats:title>\n                <jats:p>\n                  The apolipoprotein E\n                  <jats:italic>\u03b5<\/jats:italic>\n                  4 (APOE4) allele is the strongest genetic risk factor for late-onset Alzheimer\u2019s disease (AD), the most common form of dementia. APOE4 carriers exhibit cerebrovascular and metabolic dysfunction, structural brain alterations, and gut microbiome changes decades before the onset of clinical symptoms. Better understanding of the early manifestion of these physiological changes is critical for development of timely AD interventions and risk reduction protocols. Multi-modal datasets encompassing a wide range of APOE\n                  <jats:italic>\u03b5<\/jats:italic>\n                  4 and AD associated biomarkers provide a valuable opportunity to gain insight into the APOE4 phenotype; however, these datasets often present analytical challenges due to small sample sizes and high heterogeneity. Here, we propose a two-stage multimodal AI model (APOEFormer) that integrates blood metabolites, brain vascular and structural MRI, microbiome profiles, and other clinical and demographic data to predict APOE4 allele status. In the first stage, modality-specific encoders are used to generate initial representa-tions of input data modalities, which are aligned in a shared latent space via self-supervised contrastive learning during pretraining. The contrastive learning objective encourages learning of informative and consistent representations across modalities through leveraging cross-modality relationships. In the second stage, the pretrained representations are used as inputs to a multimodal transformer that integrates information across modalities to predict a key AD-risk genetic variant (APOE4). Across 10 independent experimental runs with different train\u2013validation\u2013test splits, APOEFormer predicts whether an individual carries an APOE4 allele with an average prediction accuracy of 75%, demonstrating robust performance under limited sample sizes. Post hoc perturbation analysis of the predictive model revealed valuable insights into the driving components of the APOE4 phenotype\u2014 including key blood biomarkers and brain regions strongly associated with APOE4.\n                <\/jats:p>","DOI":"10.64898\/2026.04.13.26350803","type":"posted-content","created":{"date-parts":[[2026,4,15]],"date-time":"2026-04-15T21:20:11Z","timestamp":1776288011000},"source":"Crossref","is-referenced-by-count":1,"title":["A multimodal AI model for modeling the genetic risk factor of Alzheimer\u2019s disease"],"prefix":"10.64898","author":[{"given":"Thong","family":"Nguyen","sequence":"first","affiliation":[{"name":"Department of Electrical Engineering and Computer Science, University of Missouri, Columbia, MO, USA"},{"name":"Roy Blunt NextGen Precision Health, University of Missouri, Columbia, MO, USA"}],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Carter","family":"Woods","sequence":"additional","affiliation":[{"name":"Roy Blunt NextGen Precision Health, University of Missouri, Columbia, MO, USA"}],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Jian","family":"Liu","sequence":"additional","affiliation":[{"name":"Department of Electrical Engineering and Computer Science, University of Missouri, Columbia, MO, USA"},{"name":"Roy Blunt NextGen Precision Health, University of Missouri, Columbia, MO, USA"}],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Chris","family":"Wang","sequence":"additional","affiliation":[{"name":"Department of Electrical Engineering and Computer Science, University of Missouri, Columbia, MO, USA"},{"name":"Roy Blunt NextGen Precision Health, University of Missouri, Columbia, MO, USA"}],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Ai-Ling","family":"Lin","sequence":"additional","affiliation":[{"name":"Roy Blunt NextGen Precision Health, University of Missouri, Columbia, MO, USA"}],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Jianlin","family":"Cheng","sequence":"additional","affiliation":[{"name":"Department of Electrical Engineering and Computer Science, University of Missouri, Columbia, MO, USA"},{"name":"Roy Blunt NextGen Precision Health, University of Missouri, Columbia, MO, USA"}],"role":[{"vocabulary":"crossref","role":"author"}]}],"member":"54368","reference":[{"key":"2026041717301267000_2026.04.13.26350803v1.1","doi-asserted-by":"publisher","DOI":"10.1111\/j.1582-4934.2007.00130.x"},{"issue":"5","key":"2026041717301267000_2026.04.13.26350803v1.2","doi-asserted-by":"crossref","first-page":"3708","DOI":"10.1002\/alz.13809","volume":"20","author":"\u201c2024 alzheimer\u2019s disease facts and figures,\u201d","year":"2024","journal-title":"Alzheimer\u2019s & Dementia"},{"key":"2026041717301267000_2026.04.13.26350803v1.3","doi-asserted-by":"publisher","DOI":"10.1093\/brain\/awy103"},{"issue":"1","key":"2026041717301267000_2026.04.13.26350803v1.4","doi-asserted-by":"crossref","first-page":"7","DOI":"10.1186\/s13195-023-01380-w","article-title":"Effects of apoe2 and apoe4 on brain microstructure in older adults: modification by age, sex, and cognitive status","volume":"16","year":"2024","journal-title":"Alzheimer\u2019s research & therapy"},{"issue":"3","key":"2026041717301267000_2026.04.13.26350803v1.5","doi-asserted-by":"crossref","first-page":"131","DOI":"10.1038\/nrneurol.2010.4","article-title":"Cerebrospinal fluid and plasma biomarkers in alzheimer disease","volume":"6","year":"2010","journal-title":"Nature Reviews Neurology"},{"issue":"3","key":"2026041717301267000_2026.04.13.26350803v1.6","doi-asserted-by":"crossref","first-page":"292","DOI":"10.1016\/j.jalz.2016.02.002","article-title":"Pre-clinical alzheimer\u2019s disease: definition, natural history, and diagnostic criteria","volume":"12","year":"2016","journal-title":"Alzheimer\u2019s & Dementia"},{"issue":"1","key":"2026041717301267000_2026.04.13.26350803v1.7","doi-asserted-by":"crossref","first-page":"48","DOI":"10.5056\/jnm18087","article-title":"Brain-gut-microbiota axis in alzheimer\u2019s disease","volume":"25","year":"2019","journal-title":"Journal of neurogastroenterology and motility"},{"key":"2026041717301267000_2026.04.13.26350803v1.8","doi-asserted-by":"publisher","DOI":"10.1016\/j.neuroimage.2011.01.008"},{"issue":"3","key":"2026041717301267000_2026.04.13.26350803v1.9","first-page":"584","article-title":"Alzheimer\u2019s disease diagnostics by a 3d deeply supervised adaptable convolutional network","volume":"23","year":"2018","journal-title":"Front. Biosci"},{"key":"2026041717301267000_2026.04.13.26350803v1.10","doi-asserted-by":"publisher","DOI":"10.1016\/j.neuroimage.2014.06.077"},{"key":"2026041717301267000_2026.04.13.26350803v1.11","unstructured":"T. Chen , S. Kornblith , M. Norouzi , and G. Hinton , \u201cA simple framework for contrastive learning of visual representations,\u201d in International conference on machine learning. PmLR, 2020, pp. 1597\u20131607."},{"key":"2026041717301267000_2026.04.13.26350803v1.12","unstructured":"A. Radford , J. W. Kim , C. Hallacy , A. Ramesh , G. Goh , S. Agarwal , G. Sastry , A. Askell , P. Mishkin , J. Clark et al., \u201cLearning transferable visual models from natural language supervision,\u201d in International conference on machine learning. PmLR, 2021, pp. 8748\u20138763."},{"key":"2026041717301267000_2026.04.13.26350803v1.13","doi-asserted-by":"crossref","first-page":"102656","DOI":"10.1016\/j.media.2022.102656","article-title":"Multi-modal contrastive mutual learning and pseudo-label re-learning for semi-supervised medical image segmentation","volume":"83","year":"2023","journal-title":"Medical Image Analysis"},{"key":"2026041717301267000_2026.04.13.26350803v1.14","unstructured":"A. Vaswani , N. Shazeer , N. Parmar , J. Uszkoreit , L. Jones , A. N. Gomez , \u0141. Kaiser , and I. Polosukhin , \u201cAttention is all you need,\u201d Advances in neural information processing systems, vol. 30, 2017."},{"key":"2026041717301267000_2026.04.13.26350803v1.15","unstructured":"A. Vaswani , N. Shazeer , N. Parmar , J. Uszkoreit , L. Jones , A. N. Gomez , \u0141. Kaiser , and I. Polosukhin , \u201cAttention is all you need,\u201d Advances in neural information processing systems, vol. 30, 2017."},{"key":"2026041717301267000_2026.04.13.26350803v1.16","article-title":"& houlsby, n.(2020). an image is worth 16x16 words: Transformers for image recognition at scale","year":"2010","journal-title":"arXiv preprint"},{"issue":"6","key":"2026041717301267000_2026.04.13.26350803v1.17","doi-asserted-by":"crossref","first-page":"743","DOI":"10.1038\/s41551-023-01045-x","article-title":"A transformer-based representation-learning model with unified processing of multimodal input for clinical diagnostics","volume":"7","year":"2023","journal-title":"Nature biomedical engineering"},{"key":"2026041717301267000_2026.04.13.26350803v1.18","unstructured":"A. Radford , J. W. Kim , C. Hallacy , A. Ramesh , G. Goh , S. Agarwal , G. Sastry , A. Askell , P. Mishkin , J. Clark et al., \u201cLearning transferable visual models from natural language supervision,\u201d in International conference on machine learning. PmLR, 2021, pp. 8748\u20138763."},{"issue":"5","key":"2026041717301267000_2026.04.13.26350803v1.19","first-page":"e111","article-title":"The alzheimer\u2019s disease neuroimaging initiative: a review of papers published since its inception","volume":"9","year":"2013","journal-title":"Alzheimer\u2019s & Dementia"},{"key":"2026041717301267000_2026.04.13.26350803v1.20","doi-asserted-by":"crossref","unstructured":"J. Devlin , M.-W. Chang , K. Lee , and K. Toutanova , \u201cBert: Pre-training of deep bidirectional transformers for language understanding,\u201d in Proceedings of the 2019 conference of the North American chapter of the association for computational linguistics: human language technologies, volume 1 (long and short papers), 2019, pp. 4171\u20134186.","DOI":"10.18653\/v1\/N19-1423"},{"issue":"2","key":"2026041717301267000_2026.04.13.26350803v1.21","doi-asserted-by":"crossref","first-page":"106","DOI":"10.1038\/nrneurol.2012.263","article-title":"Apolipoprotein e and alzheimer disease: risk, mechanisms and therapy","volume":"9","year":"2013","journal-title":"Nature Reviews Neurology"},{"key":"2026041717301267000_2026.04.13.26350803v1.22","doi-asserted-by":"publisher","DOI":"10.1016\/j.media.2017.07.005"},{"key":"2026041717301267000_2026.04.13.26350803v1.23","article-title":"Layer normalization","year":"2016","journal-title":"arXiv preprint"},{"key":"2026041717301267000_2026.04.13.26350803v1.24","first-page":"2127","volume-title":"Proceedings of the International Conference on Machine Learning, ser. Proceedings of Machine Learning Research","volume":"80","year":"2018"},{"key":"2026041717301267000_2026.04.13.26350803v1.25","doi-asserted-by":"crossref","unstructured":"T.-Y. Lin , P. Goyal , R. Girshick , K. He , and P. Doll\u00e1r, \u201cFocal loss for dense object detection,\u201d in Proceedings of the IEEE International Conference on Computer Vision, 2017, pp. 2980\u20132988.","DOI":"10.1109\/ICCV.2017.324"},{"issue":"1","key":"2026041717301267000_2026.04.13.26350803v1.26","doi-asserted-by":"crossref","first-page":"119","DOI":"10.1016\/S1474-4422(09)70299-6","article-title":"Hypothetical model of dynamic biomarkers of the alzheimer\u2019s pathological cascade","volume":"9","year":"2010","journal-title":"The Lancet Neurology"},{"key":"2026041717301267000_2026.04.13.26350803v1.27","doi-asserted-by":"publisher","DOI":"10.1002\/mrm.25197"},{"key":"2026041717301267000_2026.04.13.26350803v1.28","doi-asserted-by":"crossref","unstructured":"M. D. Zeiler and R. Fergus , \u201cVisualizing and understanding convolutional networks,\u201d in European conference on computer vision. Springer, 2014, pp. 818\u2013833.","DOI":"10.1007\/978-3-319-10590-1_53"},{"key":"2026041717301267000_2026.04.13.26350803v1.29","doi-asserted-by":"crossref","unstructured":"R. C. Fong and A. Vedaldi , \u201cInterpretable explanations of black boxes by meaningful perturbation,\u201d in Proceedings of the IEEE international conference on computer vision, 2017, pp. 3429\u20133437.","DOI":"10.1109\/ICCV.2017.371"},{"key":"2026041717301267000_2026.04.13.26350803v1.30","unstructured":"C. Molnar , Interpretable Machine Learning, 3rd ed., 2025. [Online]. Available: https:\/\/christophm.github.io\/interpretable-ml-book"},{"issue":"4","key":"2026041717301267000_2026.04.13.26350803v1.31","doi-asserted-by":"crossref","first-page":"388","DOI":"10.1016\/S1474-4422(15)70016-5","article-title":"Neuroinflammation in alzheimer\u2019s disease","volume":"14","year":"2015","journal-title":"The Lancet Neurology"},{"issue":"2","key":"2026041717301267000_2026.04.13.26350803v1.32","doi-asserted-by":"crossref","first-page":"67","DOI":"10.1038\/nrneurol.2009.215","article-title":"The clinical use of structural mri in alzheimer disease","volume":"6","year":"2010","journal-title":"Nature reviews neurology"},{"key":"2026041717301267000_2026.04.13.26350803v1.33","doi-asserted-by":"publisher","DOI":"10.1126\/scitranslmed.abo2984"},{"key":"2026041717301267000_2026.04.13.26350803v1.34","unstructured":"S. M. Lundberg and S.-I. Lee , \u201cA unified approach to interpreting model predictions,\u201d Advances in neural information processing systems, vol. 30, 2017."},{"key":"2026041717301267000_2026.04.13.26350803v1.35","doi-asserted-by":"publisher","DOI":"10.1073\/pnas.0308627101"},{"key":"2026041717301267000_2026.04.13.26350803v1.36","doi-asserted-by":"publisher","DOI":"10.1196\/annals.1440.011"},{"key":"2026041717301267000_2026.04.13.26350803v1.37","doi-asserted-by":"publisher","DOI":"10.1152\/physrev.00006.2011"},{"key":"2026041717301267000_2026.04.13.26350803v1.38","doi-asserted-by":"publisher","DOI":"10.1196\/annals.1427.007"},{"key":"2026041717301267000_2026.04.13.26350803v1.39","doi-asserted-by":"publisher","DOI":"10.1371\/journal.pone.0063644"},{"key":"2026041717301267000_2026.04.13.26350803v1.40","doi-asserted-by":"publisher","DOI":"10.1038\/nrn.2016.141"},{"key":"2026041717301267000_2026.04.13.26350803v1.41","doi-asserted-by":"publisher","DOI":"10.1681\/ASN.2012050536"},{"issue":"3","key":"2026041717301267000_2026.04.13.26350803v1.42","doi-asserted-by":"crossref","first-page":"1403","DOI":"10.3233\/JAD-170585","article-title":"Mitochondria and mitochondrial cascades in alzheimer\u2019s disease","volume":"62","year":"2018","journal-title":"Journal of Alzheimer\u2019s Disease"},{"issue":"1","key":"2026041717301267000_2026.04.13.26350803v1.43","doi-asserted-by":"crossref","first-page":"64","DOI":"10.4103\/ni.ni_211_22","article-title":"Red cell distribution width is related to mild cognitive impairment: A cross-sectional study of community residents","volume":"72","year":"2024","journal-title":"Neurology India"},{"issue":"4","key":"2026041717301267000_2026.04.13.26350803v1.44","doi-asserted-by":"crossref","first-page":"388","DOI":"10.1016\/S1474-4422(15)70016-5","article-title":"Neuroinflammation in alzheimer\u2019s disease","volume":"14","year":"2015","journal-title":"The Lancet Neurology"},{"key":"2026041717301267000_2026.04.13.26350803v1.45","first-page":"575","article-title":"Inflammation as a central mechanism in alzheimer\u2019s disease","volume":"4","year":"2018","journal-title":"Alzheimer\u2019s & Dementia: Translational Research & Clinical Interventions"},{"key":"2026041717301267000_2026.04.13.26350803v1.46","unstructured":"J. Lee , Y. Lee , J. Kim , A. Kosiorek , S. Choi , and Y. W. Teh , \u201cSet transformer: A framework for attention-based permutation-invariant neural networks,\u201d in International conference on machine learning. PMLR, 2019, pp. 3744\u20133753."},{"key":"2026041717301267000_2026.04.13.26350803v1.47","doi-asserted-by":"publisher","DOI":"10.1016\/j.neuroimage.2017.06.061"},{"key":"2026041717301267000_2026.04.13.26350803v1.48","doi-asserted-by":"publisher","DOI":"10.1038\/nn.4478"},{"issue":"9","key":"2026041717301267000_2026.04.13.26350803v1.49","doi-asserted-by":"crossref","first-page":"501","DOI":"10.1038\/s41582-019-0228-7","article-title":"Apolipoprotein e and alzheimer disease: Pathobiology and targeting strategies","volume":"15","year":"2019","journal-title":"Nature Reviews Neurology"}],"container-title":[],"original-title":[],"link":[{"URL":"https:\/\/syndication.highwire.org\/content\/doi\/10.64898\/2026.04.13.26350803","content-type":"unspecified","content-version":"vor","intended-application":"similarity-checking"}],"deposited":{"date-parts":[[2026,4,18]],"date-time":"2026-04-18T00:30:23Z","timestamp":1776472223000},"score":1,"resource":{"primary":{"URL":"http:\/\/medrxiv.org\/lookup\/doi\/10.64898\/2026.04.13.26350803"}},"subtitle":[],"short-title":[],"issued":{"date-parts":[[2026,4,15]]},"references-count":49,"URL":"https:\/\/doi.org\/10.64898\/2026.04.13.26350803","relation":{},"subject":[],"published":{"date-parts":[[2026,4,15]]},"subtype":"preprint"}}