FEDERATED MULTIMODAL GRAPH
NEURAL NETWORK FOR BIAS-AWARE
EARLY DETECTION OF
PANCREATIC CANCER
Abul Walid
Director & Chief Technology Officer,
Vimo Software Development Private Limited,
Bengaluru, India
Abstract. The timely diagnosis of pancreatic cancer has been a critical clinical issue because of insensitive early disease manifestations, heterogeneity of data, and rigid privacy requirements. This paper suggests Federated Multimodal Graph Neural Network (FM-GNN) architecture involving the use of CT images and clinical biomarkers to diagnose pancreatic cancer at earlier stages without bias. Multimodal features are combined and represented in the form of a patient similarity graph, which allows the relational learning of graph neural networks. Federated learning enables joint training in simulated healthcare institutions without raw data sharing. The experimental outcomes indicate stable federated convergence, competitive ROC-AUC performance, and effective detection of the cancer, which indicates the possibility of the framework in the privacy-preserving and fair application in clinical practice.
How
to cite this paper?
Source: International Journal of Applied Mathematics
ISSN printed version: 1311-1728
ISSN on-line version: 1314-8060
Year: 2025
Volume: 38
Issue: 4
References
[1] Tiwari, P., Kumar, M.V., Kadiri, S.K., Rathee, S., Soni, S. and Sen, D., 2026. Precision treatment planning and decision support systems. Artificial Intelligence-Driven Precision Medicine for Triple Negative Breast Cancer, pp.227-241.
[2] Kazakova, A.N., Anufrieva, K.S., Ivanova, O.M., Shnaider, P.V., Malyants, I.K., Aleshikova, O.I., Slonov, A.V., Ashrafyan, L.A., Babaeva, N.A., Eremeev, A.V. and Boichenko, V.S., 2022. Deeper insights into transcriptional features of cancer-associated fibroblasts: An integrated meta-analysis of single-cell and bulk RNA-sequencing data. Frontiers in Cell and Developmental Biology, 10, p.825014.
[3] Kazakova, A.N., Anufrieva, K.S., Ivanova, O.M., Shnaider, P.V., Malyants, I.K., Aleshikova, O.I., Slonov, A.V., Ashrafyan, L.A., Babaeva, N.A., Eremeev, A.V. and Boichenko, V.S., 2022. Deeper insights into transcriptional features of cancer-associated fibroblasts: An integrated meta-analysis of single-cell and bulk RNA-sequencing data. Frontiers in Cell and Developmental Biology, 10, p.825014.
[4] Attema, B.J., 2024. Exposing metabolic health: Insights in the effects of environmental chemicals, nuclear receptors and SLC25A47 on hepatic metabolism (Doctoral dissertation, Wageningen University and Research).
[5] Chicco, D. and Jurman, G., 2023. The Matthews correlation coefficient (MCC) should replace the ROC AUC as the standard metric for assessing binary classification. BioData Mining, 16(1), p.4.
[6] Carrington, A.M., Manuel, D.G., Fieguth, P.W., Ramsay, T., Osmani, V., Wernly, B., Bennett, C., Hawken, S., Magwood, O., Sheikh, Y. and McInnes, M., 2022. Deep ROC analysis and AUC as balanced average accuracy, for improved classifier selection, audit and explanation. IEEE Transactions on Pattern Analysis and Machine Intelligence, 45(1), pp.329-341.
[7] Li, J., 2024. Area under the ROC Curve has the most consistent evaluation for binary classification. PloS one, 19(12), p.e0316019.
[8] Crimì, F., D’Alessandro, C., Zanon, C., Celotto, F., Salvatore, C., Interlenghi, M., Castiglioni, I., Quaia, E., Pucciarelli, S. and Spolverato, G., 2024. A Machine Learning Model Based on MRI Radiomics to Predict Response to Chemoradiation Among Patients with Rectal Cancer. Life, 14(12), p.1530.
[9] Liu, J., Wang, Y., Lv, W., Gao, Z., Su, S., Guo, B., Li, Y., Liang, L., Wang, B., Cheng, G. and Zhang, J., 2025. Ultra-early detection of S100B biomarkers using a nanophotonic biosensor with deep learning quantification: A clinical model based on EDAS patients. Biosensors and Bioelectronics, p.117810.
[10] Hong, J., Hachem, L.D. and Fehlings, M.G., 2022. A deep learning model to classify neoplastic state and tissue origin from transcriptomic data. Scientific reports, 12(1), p.9669.
[11] Sofiia, V., Ilya, D., Levon, A. and Aleksandra, V., 2024, June. Graph-Based Data Representation and Prediction in Medical Domain Tasks Using Graph Neural Networks. In International Conference on Computational Science (pp. 371-378). Cham: Springer Nature Switzerland.
[12] Outemzabet, L., Gaud, N., Bertaux, A., Nicolle, C., Gerart, S., Vachenc, S., Riveiro, M.E., Malliaros, F.D. and Raieli, S., 2025. From Predicting Cancer Treatment Response to Identifying Novel Therapeutic Targets using Graph Neural Networks. IEEE Journal of Biomedical and Health Informatics.
[13] Zi, S., Borde, H.S.D.O., Rocheteau, E. and Liò, P., 2025. Bridging Graph and State-Space Modeling for Intensive Care Unit Length of Stay Prediction. arXiv preprint arXiv:2508.17554.
[14] David, S., 2025. Graph Neural Networks for Modeling Patient Relationships in Early Disease Progression Prediction.
[15] Molaei, S., Bousejin, N.G., Ghosheh, G.O., Thakur, A., Chauhan, V.K., Zhu, T. and Clifton, D.A., 2024. CliqueFluxNet: unveiling EHR insights with stochastic edge fluxing and maximal clique utilisation using graph neural networks. Journal of Healthcare Informatics Research, 8(3), pp.555-575.
[16] Berkani, K., 2025. APC-GNN++: An Adaptive Patient-Centric GNN with Context-Aware Attention and Mini-Graph Explainability for Diabetes Classification. arXiv preprint arXiv:2512.18473.
[17] Ornithopoulou, E., 2025. FairGATDANN: A Fairness-Aware Graph Attention and Domain-Adversarial Neural Network for ICU Mortality Prediction.
[18] Issaiyan, N., Abdollahpouri, A., Sulaimany, S. and Dibaji, A., Graph Neural Networks for Latent Feature Learning Beyond Tabular Data: Application to Thyroid Disease Diagnosis. Available at SSRN 5295704.
[19] Lu, M. and Lu, H., 2025. A hypergraph neural network based framework for Identification of Parkinson’s disease subtypes. Chinese Physics B.
[20] Shivasree, Y. and RaviSankar, V., 2025. Design of an iterative hybrid multimodal deep learning method for early skin disease detection with cross-attention and graph-based fusions. MethodsX, p.103584.
[21] Hu, H., Liang, W., Zou, X. and Zou, X., 2024, July. Graph Convolutional Networks Based Multi-modal Data Integration for Breast Cancer Survival Prediction. In International Conference on Intelligent Computing (pp. 85-98). Singapore: Springer Nature Singapore.
[22] Yang, W., Yoshida, S., Zhao, J., Wu, W. and Qiang, Y., 2025. MVASA-HGN: multi-view adaptive semantic-aware heterogeneous graph network for KRAS mutation status prediction. Quantitative Imaging in Medicine and Surgery, 15(2), p.1190.
[23] Wu, Z., Dadu, A., Tustison, N., Avants, B., Nalls, M., Sun, J. and Faghri, F., 2024. Multimodal patient representation learning with missing modalities and labels. In The Twelfth International Conference on Learning Representations.
[24] Yang, Z., Li, K., Gan, H., Huang, Z., Shi, M. and Zhou, R., 2024. An Alzheimer’s disease classification network based on MRI utilizing diffusion maps for multi-scale feature fusion in graph convolution. Math Biosci Eng, 21(1), pp.1554-1572.
[25] Fan, G., Liu, A. and Zhang, C., 2025. ContrastLOS: A Graph-Based Deep Learning Model With Contrastive Pre-Training for Improved ICU Length-of-Stay Prediction. IEEE Access.
[26] Hou, M., Xia, Y., Wang, P., Lv, Z., Hou, D., Zhou, X., Zeng, J. and Zhang, G., 2025. High-accuracy protein complex structure modeling based on sequence-derived structure complementarity. Nature Communications, 16(1), p.10182.
[27] Kanoun, Y., Aghbash, A.M., Belem, T., Zouari, B. and Mrad, H., 2024. Failure prediction in the refinery piping system using machine learning algorithms: Classification and comparison. Procedia Computer Science, 232, pp.1663-1672.
[28] Dhivya, P., Rajesh Kanna, P., Deepa, K. and Santhiya, S., 2024. Square static–deep hyper optimization and genetic meta-learning approach for disease classification. IETE Journal of Research, 70(4), pp.3835-3844.
[29] Nijhawan, R., Kumar, M., Arya, S., Mendirtta, N., Kumar, S., Towfek, S.K., Khafaga, D.S., Alkahtani, H.K. and Abdelhamid, A.A., 2023. A novel artificial-intelligence-based approach for classification of Parkinson’s disease using complex and large vocal features. Biomimetics, 8(4), p.351.
[30] Veasey, B.P. and Amini, A.A., 2025. Low-rank adaptation of pre-trained large vision models for improved lung nodule malignancy classification. IEEE Open Journal of Engineering in Medicine and Biology.
[31] Ghourabi, A., 2022. A security model based on lightgbm and transformer to protect healthcare systems from cyberattacks. IEEE Access, 10, pp.48890-48903.
[32] Noh, S.H., 2023. Comparing the performance of corporate bankruptcy prediction models based on imbalanced financial data. Sustainability, 15(6), p.4794.
IJAM
o Home
o Contents