Multimodal deep learning model for prediction of prognosis in central nervous system inflammation
Open accessThe DOI matches its Crossref record; 37 of the 37 references it cites were found in a registry or an index.
Abstract
Inflammatory diseases of the CNS impose a substantial disease burden, necessitating prompt and appropriate prognosis prediction. We developed a multimodal deep learning model integrating clinical features and brain MRI data to enhance early prognosis prediction of CNS inflammation. This retrospective study used thin-cut T1-weighted brain MRI scans and the clinical variables of patients with CNS inflammation who were admitted to a tertiary referral hospital between January 2010 and December 2023. Data collected after January 2022 served as the external test set. 3D MRI images were first segmented into 43 brain regions using the FastSurfer library. The segmented images were then processed through a 3D convolutional neural network model for feature extraction and vectorization, after which they were integrated with clinical features for prediction. The performance of each artificial intelligence model was assessed using accuracy, F1 score, area under the receiver operating characteristic curve and area under the precision-recall curve. The internal dataset comprised 413 images from 291 patients (mean age, 45.5 years ± 19.3 [SD]; 151 male patients; 54 with poor prognosis). The external dataset comprised 210 images from 106 patients (mean age, 45.5 years ± 18.9 [SD]; 59 male patients; 31 with poor prognosis). The multimodal deep learning model outperformed unimodal models across all aetiological groups, achieving area under the receiver operating characteristic curve values of 0.8048 for autoimmune, 0.9107 for bacterial, 1.0000 for tuberculosis and 0.9242 for viral infections. Furthermore, artificial intelligence assistance improved clinicians' prognostic accuracy, as demonstrated in comparisons with neurologists, paediatricians and radiologists. Our findings demonstrate that the multimodal deep learning model enhances artificial intelligence-assisted prognosis prediction in CNS inflammation, improving both model performance and clinician decision-making.
Overview
The claims and statistics of this paper haven’t been checked yet.
Free: it reads the open-access text and checks the first 5 claims that cite a source we can check. Create a free account to check every claim.
Integrity signals
Registry
Cites
Statistics
Figures




Verification
Claims the paper makes
The claims and statistics of this paper haven’t been checked yet.
Free: it reads the open-access text and checks the first 5 claims that cite a source we can check. Create a free account to check every claim.
Self-check
Percentages, sums and reported differences re-computed from the paper’s own numbers.
The self-check needs the paper’s full text.
Citations
Cites
A Unified Approach to Interpreting Model Predictions
Lundberg, S., & Lee, S.-I. (2017) · arXiv · doi:10.48550/arxiv.1705.07874 · Cited by 23,707 · cited 1×
“16 As SHAP values are model agnostic, it can be used to explain contributions of features across different modalities.”
Gauriat, C.-M., Pencolé, Y., Ribot, P., & Brouillet, G. (2024) · HAL (Le Centre pour la Communication Scientifique Directe) · doi:10.4230/oasics.dx.2024.27 · Cited by 12,586
Explaining prediction models and individual predictions with feature contributions
Štrumbelj, E., & Kononenko, I. (2014) · Knowledge and Information Systems · doi:10.1007/s10115-013-0679-x · Cited by 2,286 · cited 1×
“The approximation method was implemented using coalition vectors 18 and the details of the approximation algorithm are provided in Appendix 1 of the Supplementary Materials.”
Acosta, J. N., Falcone, G. J., Rajpurkar, P., & Topol, E. J. (2022) · Nature Medicine · doi:10.1038/s41591-022-01981-2 · Cited by 1,365
Granerod, J., Ambrose, H. E., Davies, N. W., Clewley, J. P., Walsh, A. L., Morgan, D., Cunningham, R., Zuckerman, M., Mutton, K. J., Solomon, T., et al. (2010) · The Lancet Infectious Diseases · doi:10.1016/s1473-3099(10)70222-x · Cited by 1,358 · cited 1×
“2,3 However, due to economic and human resource constraints, it is necessary to predict and prepare for disease aggravation in advance, as continuous monitoring for all patients is impractical.”
Huang, S.-C., Pareek, A., Seyyedi, S., Banerjee, I., & Lungren, M. P. (2020) · npj Digital Medicine · doi:10.1038/s41746-020-00341-z · Cited by 934 · cited 1×
“23 The brain MRI-based unimodal model outperformed the clinical variable-based model in predicting outcomes for viral infections.”
Multi-Classification of Brain Tumor Images Using Deep Neural Network
Sultan, H. H., Salem, N. M., & Al-Atabany, W. (2019) · IEEE Access · doi:10.1109/access.2019.2919122 · Cited by 775 · cited 2×
“4,8,9,24 This method was mainly applied to neurodegenerative diseases with overall brain atrophy or relatively large and clearly visible brain tumours, aiming to reduce information loss T a b l e 1 and allow intuitive analysis.”
Manner, J., Haarmann, S., Kolb, S., Herzberg, N., & Kopp, O. (2021) · Cited by 732
FastSurfer - A fast and accurate deep learning based neuroimaging pipeline
Henschel, L., Conjeti, S., Estrada, S., Diers, K., Fischl, B., & Reuter, M. (2020) · NeuroImage · doi:10.1016/j.neuroimage.2020.117012 · Cited by 660
Autoimmune encephalitis: proposed best practice recommendations for diagnosis and acute management
Abboud, H., Probasco, J. C., Irani, S., Ances, B., Benavides, D. R., Bradshaw, M., Christo, P. P., Dale, R. C., Fernandez-Fournier, M., Flanagan, E. P., et al. (2021) · Journal of Neurology, Neurosurgery & Psychiatry · doi:10.1136/jnnp-2020-325300 · Cited by 554 · cited 2×
“28 Hence, the final predictive performance is presumed to be the most unsatisfactory.”
Limbic Encephalitis and Variants: Classification, Diagnosis and Treatment
Tüzün, E., & Dalmau, J. (2007) · The Neurologist · doi:10.1097/nrl.0b013e31813e34a5 · Cited by 478 · cited 1×
“26,28,32 Consistently, in our study, these regions emerged as important predictors for prognosis.”
Solomon, T., Michael, B. D., Smith, P. E., Sanderson, F., Davies, N. W. S., Hart, I. J., Holland, M., Easton, A., Buckley, C., Kneen, R., et al. (2012) · Journal of Infection · doi:10.1016/j.jinf.2011.11.014 · Cited by 464 · cited 1×
“2,3 However, due to economic and human resource constraints, it is necessary to predict and prepare for disease aggravation in advance, as continuous monitoring for all patients is impractical.”
Multimodal deep learning for Alzheimer's disease dementia assessment
Qiu, S., Miller, M. I., Joshi, P. S., Lee, J. C., Xue, C., Ni, Y., Wang, Y., De Anda-Duran, I., Hwang, P. H., Cramer, J. A., et al. (2022) · Nature Communications · doi:10.1038/s41467-022-31037-5 · Cited by 420 · cited 2×
“4,8,9,24 This method was mainly applied to neurodegenerative diseases with overall brain atrophy or relatively large and clearly visible brain tumours, aiming to reduce information loss T a b l e 1 and allow intuitive analysis.”
Zunt, J. R., Kassebaum, N. J., Blake, N., Glennie, L., Wright, C., Nichols, E., Abd-Allah, F., Abdela, J., Abdelalim, A., Adamu, A. A., et al. (2018) · The Lancet Neurology · doi:10.1016/s1474-4422(18)30387-9 · Cited by 385
Community-acquired bacterial meningitis
van de Beek, D., Brouwer, M., Hasbun, R., Koedel, U., Whitney, C. G., & Wijdicks, E. (2016) · Nature Reviews Disease Primers · doi:10.1038/nrdp.2016.74 · Cited by 312 · cited 1×
“33 This underscores the importance of paying close attention to CSF test results in clinical practice.”
Viral meningoencephalitis: a review of diagnostic methods and guidelines for management
Steiner, I., Budka, H., Chaudhuri, A., Koskiniemi, M., Sainio, K., Salonen, O., & Kennedy, P. G. E. (2010) · European Journal of Neurology · doi:10.1111/j.1468-1331.2010.02970.x · Cited by 238 · cited 2×
“26,27 Based on this, performing brain MRI scans when viral brain infection is suspected could be particularly helpful in prognostic prediction.”
Shaik, T., Tao, X., Li, L., Xie, H., & Velásquez, J. D. (2024) · Information Fusion · doi:10.1016/j.inffus.2023.102040 · Cited by 196 · cited 1×
“29 The clinical variables selected for MMDL differed depending on the aetiology.”
Wang, Y.-R., Yang, K., Wen, Y., Wang, P., Hu, Y., Lai, Y., Wang, Y., Zhao, K., Tang, S., Zhang, A., et al. (2024) · Nature Medicine · doi:10.1038/s41591-024-02971-2 · Cited by 188
Prognosticating autoimmune encephalitis: A systematic review
Broadley, J., Seneviratne, U., Beech, P., Buzzard, K., Butzkueven, H., O'Brien, T., & Monif, M. (2019) · Journal of Autoimmunity · doi:10.1016/j.jaut.2018.10.014 · Cited by 183 · cited 1×
“21 re-ported a review article analysing studies that predicted the prognosis of autoimmune encephalitis using clinical variables.”
Patel, K., & Clifford, D. B. (2014) · The Neurohospitalist · doi:10.1177/1941874414540684 · Cited by 171 · cited 1×
“34 Although previous studies reported no difference in prognosis based on the location of brain abscesses, 35 our findings suggest that changes in the occipital lobe should also be considered significant.”
Central Nervous System Infections of Herpesvirus Family
Bulakbaşı, N., & Kocaoğlu, M. (2008) · Neuroimaging Clinics of North America · doi:10.1016/j.nic.2007.12.001 · Cited by 116 · cited 1×
“36 The high SHAP value in autoimmune cases was not confined to specific brain regions.”
Electroencephalography for diagnosis and prognosis of acute encephalitis
Sutter, R., Kaplan, P. W., Cervenka, M. C., Thakur, K. T., Asemota, A. O., Venkatesan, A., & Geocadin, R. G. (2014) · Clinical Neurophysiology · doi:10.1016/j.clinph.2014.11.006 · Cited by 97 · cited 1×
“7,22 While the model based solely on brain MRI modality did not show high predictive performance, the multimodal model combining both modalities demonstrated good predictive performance.”
A Learning Strategy for Contrast-agnostic MRI SegmentationPreprint
Billot, B., Greve, D. N., Leemput, K. V., Fischl, B. R., Iglesias, J. E., & Dalca, A. V. (2020) · arXiv (Cornell University) · doi:10.48550/arxiv.2003.01995 · Cited by 92
Wu, Q., Wang, S., Zhang, S., Wang, M., Ding, Y., Jin, F., Wu, Q., Qian, W., Liu, Z., Sun, K., et al. (2020) · JAMA Network Open · doi:10.1001/jamanetworkopen.2020.11625 · Cited by 90 · cited 2×
“12 suggested a MMDL model using preoperative MRI to noninvasively predict lymph node metastasis in cervical cancer demonstrating prognostic value for disease-free survival.”
Magnetic Resonance Imaging Findings in Viral Encephalitis: A Pictorial Essay
Jayaraman, K., Rangasami, R., & Chandrasekharan, A. (2018) · Journal of Neurosciences in Rural Practice · doi:10.4103/jnrp.jnrp_120_18 · Cited by 85 · cited 1×
“26,27 Based on this, performing brain MRI scans when viral brain infection is suspected could be particularly helpful in prognostic prediction.”
Neuroimaging in encephalitis: analysis of imaging findings and interobserver agreement
Granerød, J., Davies, N. W. S., Mukonoweshuro, W., Mehta, A., Das, K., Lim, M. J., Solomon, T., Biswas, S., Rosella, L. C., Brown, D., et al. (2016) · Clinical Radiology · doi:10.1016/j.crad.2016.03.015 · Cited by 78 · cited 1×
“25 To create a model that better reflects real clinical situations, we selected brain areas for use in MMDL through a grouping process.”
Clinical characteristics and outcome of primary brain abscess: a retrospective analysis
Huang, J., Wu, H., Huang, H., Wu, W., Wu, B., & Wang, L. (2021) · BMC Infectious Diseases · doi:10.1186/s12879-021-06947-2 · Cited by 47 · cited 1×
“34 Although previous studies reported no difference in prognosis based on the location of brain abscesses, 35 our findings suggest that changes in the occipital lobe should also be considered significant.”
Zhou, B., Li, Y., & Wang, J. (2018) · arXiv (Cornell University) · doi:10.48550/arxiv.1807.01257 · Cited by 25 · cited 1×
“15 The training was performed with 4-fold cross validation and the model with the best performance was chosen for the inference task (Supplementary Fig.”
Severe, self-limiting lactic acidosis and rhabdomyolysis accompanying convulsions
Winocour, P. H., Waise, A., Young, G., & Moriarty, K. J. (1989) · Postgraduate Medical Journal · doi:10.1136/pgmj.65.763.321 · Cited by 22
Deep Learning‐Enabled Identification of Autoimmune Encephalitis on 3D Multi‐Sequence MRI
Xiang, Y., Zeng, C., Liu, B., Tan, W., Wu, J., Hu, X., Han, Y., Luo, Q., Gong, J., Liu, J., et al. (2021) · Journal of Magnetic Resonance Imaging · doi:10.1002/jmri.27909 · Cited by 21 · cited 1×
“4,8,9,24 This method was mainly applied to neurodegenerative diseases with overall brain atrophy or relatively large and clearly visible brain tumours, aiming to reduce information loss T a b l e 1 and allow intuitive analysis.”
Xiang, Y., Dong, X., Zeng, C., Liu, J., Liu, H., Hu, X., Feng, J., Du, S., Wang, J., Han, Y., et al. (2022) · Frontiers in Immunology · doi:10.3389/fimmu.2022.913703 · Cited by 20 · cited 2×
“4,8,9,24 This method was mainly applied to neurodegenerative diseases with overall brain atrophy or relatively large and clearly visible brain tumours, aiming to reduce information loss T a b l e 1 and allow intuitive analysis.”
Choi, B. K., Choi, Y. J., Sung, M., Ha, W., Chu, M. K., Kim, W., Heo, K., Kim, K. M., & Park, Y. R. (2023) · EClinicalMedicine · doi:10.1016/j.eclinm.2023.102051 · Cited by 20 · cited 2×
“20 Also, Broadley et al .”
Self-citationProcalcitonin as a potential predicting factor for prognosis in bacterial meningitis
Park, B. S., Kim, S. E., Park, S. H., Kim, J., Shin, K. J., Ha, S. Y., Park, J., Kim, S. E., Lee, B. I., & Park, K. M. (2016) · Journal of Clinical Neuroscience · doi:10.1016/j.jocn.2016.10.005 · Cited by 17 · cited 1×
“30 Overall, the level of consciousness at admission was crucially utilized, which is consistent with clinical prior knowledge.”
Lu, G., Liu, Y., Huang, Y., Ding, J., Zeng, Q., Zhao, L., Li, M., Yu, H., & Li, Y. (2023) · Journal of Hospital Infection · doi:10.1016/j.jhin.2023.04.004 · Cited by 11 · cited 2×
“6 attempted to predict prognosis in patients with severe traumatic brain injury after craniotomy under the specific condition of infection, employing conventional machine learning techniques.”
Zhang, Y.-F., Zeng, Q., Fang, Y., Wang, W., & Chen, Y. (2022) · Therapeutic Innovation & Regulatory Science · doi:10.1007/s43441-022-00403-2 · Cited by 8
ter Horst, L., van Zeggeren, I. E., Olie, S. E., I-PACE Study Group, Brenner, J., Citroen, J., van Geel, B. M., Heckenberg, S. G. B., Jellema, K., Kester, M. I., et al. (2023) · Scientific Reports · doi:10.1038/s41598-023-48472-z · Cited by 7
Diagnosing infection of the CNS with MRI
Cuvinciuc, V., Vargas, M. I., Lövblad, K., & Haller, S. (2011) · Imaging in Medicine · doi:10.2217/iim.11.51 · Cited by 6 · cited 3×
“7,22 While the model based solely on brain MRI modality did not show high predictive performance, the multimodal model combining both modalities demonstrated good predictive performance.”
Research works printed but not linked: 1
- Ester, 1996, Proceedings—2nd international conference on knowledge discovery and data mining, KDD 1996, 226
Cited by
Camastra, C., Pelagi, A., Quattrone, A., & Sarica, A. (2026) · Brain Sciences · doi:10.3390/brainsci16040405 · Cited by 1
Metadata
- Authors
- Choi, B. K., Choi, Y., Jang, S., Ha, W.-S., Cho, S., Chang, K., Sohn, B., Kim, K. M., & Park, Y. R.
- Affiliations
- Yonsei University (KR)
- Eulji University (KR)
- Samsung Medical Center (KR)
- Corresponding author
- Kim, Kyung Min
- Park, Yu Rang
- Venue
- Brain Communications
- Publisher
- Oxford University Press (OUP)
- Published
- 2025
- Volume
- 7
- Issue
- 3
- Pages
- fcaf179-fcaf179
- ISSN
- 2632-1297 (ISSN-L)
- PMC
- PMC12082089
- OpenAlex
- W4410225724
- Cited by
- 1
- References
- 35
- Citations by year
- 2026: 1
- FWCI
- 0.40
- Type
- journal article
- Language
- en
- Open access
- gold
- doi.org
- Locations
- Brain Communications (publishedVersion, cc-by)
- PubMed Central (publishedVersion, cc-by)
- PubMed (publishedVersion)
- Funders
- Hyundai Motor Chung Mong-Koo Foundation
- Korea Health Industry Development Institute
- Topics
- Physical Sciences › Computer Science › Artificial Intelligence › Machine Learning in Healthcare
- Health Sciences › Medicine › Health Informatics › Artificial Intelligence in Healthcare and Education
- Health Sciences › Medicine › Radiology, Nuclear Medicine and Imaging › Radiomics and Machine Learning in Medical Imaging
- Keywords
- neuroinflammation, outcome prediction, multimodal deep learning, brain MRI, T1-weighted imaging, brain structure segmentation, 3D convolutional neural networks, clinical features, autoimmune encephalitis, bacterial meningitis, tuberculous meningitis, viral encephalitis, AI-assisted diagnosis, clinician decision-making
About these checks
Registry, retraction, Cites and statistics results are computed, not judged, with one exception: a printed reference that no index lists is read once by a model before it counts as a citation error. Claims are extracted by a model and kept only when their quote appears verbatim in the text; each cited source is then checked, with a model’s help, for whether it says what the claim relies on. Those verdicts can be wrong — read the claim’s quote and the cited source before relying on one.
Statistics and arithmetic are recomputed, not judged.
Confidence bands are shown only for fields calibrated on our annotated test set. None is calibrated yet, so no band is shown.
What counts as a claim ›How we measure confidence ›How we check this ›
Are you an author, or do you see a problem with this page? Request a review
Data sources
Bibliographic data: Semantic Scholar, OpenAlex, Crossref.
Contains information from the Semantic Scholar Academic Graph Datasets, made available under the ODC Attribution License (ODC-BY 1.0).
Semantic Scholar’s data is described in: Kinney, R., et al. (2023). The Semantic Scholar Open Data Platform. arXiv:2301.10140.