Explainable Multi-Omics Transformer Risk Stratification in Lung Adenocarcinoma: External Validation, Calibration, and Patient-Level Genomic Attribution

References

[1] Argelaguet, R. et al. (2020) 'MOFA+: a statistical framework for comprehensive integration of multi-modal single-cell data', Genome Biology, 21, 111. doi: 10.1186/s13059-020-02015-1.Campbell, J.D. et al. (2018) 'Multi-omics profiling of lung cancer reveals clinically relevant subtypes', Cell Reports, 23(1), pp. 184-201. doi: 10.1016/j.celrep.2018.03.033.Cancer Genome Atlas Research Network (2014) 'Comprehensive molecular profiling of lung adenocarcinoma', Nature, 511(7511), pp. 543-550. doi: 10.1038/nature13385.

[2] Cerami, E. et al. (2012) 'The cBio cancer genomics portal: an open platform for exploring multidimensional cancer genomics data', Cancer Discovery, 2(5), pp. 401-404. doi: 10.1158/2159-8290.CD-12-0095.

[3] Chingezi, E., Chingezi, L., Ganyani, L., Yelduora, P.G. and Mupa, M.N. (2026) 'Reducing repeat findings and control failures in operationally intensive U.S. enterprises: A data-driven internal audit framework for remediation, loss prevention, and governance improvement', World Journal of Advanced Research and Reviews, 30(03), pp. 2188-2199.

[4] Choga, C.R. (2026) Integrating DNA and RNA Multi-Omics for Improved Survival Prediction in Lung Cancer. Master's thesis. Yeshiva University, Department of Mathematics.

[5] Choga, C.R. and Yu, S. (2026) 'A transformer-based multi-omics clinical decision support system for genomic risk stratification in lung adenocarcinoma', manuscript draft.

[6] CoFormerSurv Consortium (2026) 'CoFormerSurv: collaborative transformer for multi-omics survival analysis', PLOS Computational Biology.

[7] Collins, G.S. et al. (2024) 'TRIPOD+AI statement: updated guidance for reporting clinical prediction models that use regression or machine learning methods', BMJ, 385, e078378. doi: 10.1136/bmj-2023-078378.

[8] Cox, D.R. (1972) 'Regression models and life-tables', Journal of the Royal Statistical Society: Series B, 34(2), pp. 187-220.

[9] FDA (2026) Clinical Decision Support Software: Guidance for Industry and Food and Drug Administration Staff. Silver Spring, MD: U.S. Food and Drug Administration.

[10] FDA, Health Canada and MHRA (2025) Good Machine Learning Practice for Medical Device Development: Guiding Principles. Silver Spring, MD: U.S. Food and Drug Administration.

[11] Gao, J. et al. (2013) 'Integrative analysis of complex cancer genomics and clinical profiles using the cBioPortal', Science Signaling, 6(269), pl1. doi: 10.1126/scisignal.2004088.

[12] Hao, Y. et al. (2020) 'Integrated multi-omics analysis for cancer prognosis', Nature Communications, 11, pp. 1-12.

[13] Heagerty, P.J. and Zheng, Y. (2005) 'Survival model predictive accuracy and ROC curves', Biometrics, 61(1), pp. 92-105.

[14] Katzman, J.L. et al. (2018) 'DeepSurv: personalized treatment recommender system using a Cox proportional hazards deep neural network', BMC Medical Research Methodology, 18, 24. doi: 10.1186/s12874-018-0482-1.

[15] Lundberg, S.M. and Lee, S.-I. (2017) 'A unified approach to interpreting model predictions', Advances in Neural Information Processing Systems, 30.

[16] Mo, Q. et al. (2013) 'iClusterPlus: integrative clustering of multiple genomic data types', Nature Methods, 10(7), pp. 679-684. doi: 10.1038/nmeth.2651.

[17] Nhova, M., Mupa, M.N., Tsveta, L., Chingezi, L. and others (2026) 'Enterprise risk management in health coverage operations: Integrating premium billing controls-first analytics framework for eligibility-to-invoice integrity', World Journal of Advanced Research and Reviews.

[18] Okebu, D.T.C. and Mupa, M.N. (2026) 'Predictive training analytics for competency-based pilot instruction: Early detection of checkride risk, procedural noncompliance and safety-relevant learning gaps', World Journal of Advanced Research and Reviews.

[19] Pawlik, P. et al. (2025) 'Clone copy number diversity is linked to survival in lung cancer', Nature, 646(8083), pp. 190-197. doi: 10.1038/s41586-025-09398-w.

[20] Rappoport, N. and Shamir, R. (2019) 'Multi-omics integration: a comparative study', Genome Biology, 20, pp. 1-14.

[21] Sundararajan, M., Taly, A. and Yan, Q. (2017) 'Axiomatic attribution for deep networks', Proceedings of the 34th International Conference on Machine Learning, pp. 3319-3328.

[22] TCGA-LUAD Kaggle mirror (n.d.) TCGA-LUAD Dataset. Kaggle. Available at: https://www.kaggle.com/datasets/nahin333/tcgaluad (Accessed: 30 July 2026).

[23] Uno, H. et al. (2007) 'Evaluating prediction rules for survival endpoints', Statistics in Medicine, 26(22), pp. 3886-3904.

[24] Vaswani, A. et al. (2017) 'Attention is all you need', Advances in Neural Information Processing Systems, 30.

[25] Vickers, A.J. and Elkin, E.B. (2006) 'Decision curve analysis: a novel method for evaluating prediction models', Medical Decision Making, 26(6), pp. 565-574.

[26] Zehir, A. et al. (2017) 'Mutational landscape of metastatic cancer revealed from prospective clinical sequencing of 10,000 patients', Nature Medicine, 23, pp. 703-713. doi: 10.1038/nm.4333.

[27] Zhang, J. et al. (2021) 'Deep learning-based multi-omics integration for survival prediction in cancer', Nature Machine Intelligence, 3(12), pp. 1039-1048. doi: 10.1038/s42256-021-00427-8.

@article{1722517,
      author = {Cleopas Russell Choga, Manyara Sandra Kasanhayi, Nkosana Mkandla, Marlon Munjoma, Munashe Naphtali Mupa},
      title = {Explainable Multi-Omics Transformer Risk Stratification in Lung Adenocarcinoma: External Validation, Calibration, and Patient-Level Genomic Attribution},
      journal = {Iconic Research And Engineering Journals},
      year = {2026},
      volume = {10},
      number = {2},
      pages = {2728-2740},
      issn = {2456-8880},
      url = {https://www.irejournals.com/formatedpaper/1722517.pdf},
      abstract = {Lung adenocarcinoma (LUAD) remains biologically heterogeneous and clinically difficult to stratify with a single data modality. Transformer architectures can, in principle, model cross-modality interactions among RNA expression, somatic mutation, copy-number variation (CNV), and clinical variables, but their clinical value depends on reproducible validation, calibration, and patient-level explainability rather than architectural novelty alone. This article develops a revised validation and reporting framework for an explainable multi-omics transformer risk-stratification model in LUAD. The study aligns TCGA-LUAD development results, complete-case multi-omics survival modelling, and MSK-IMPACT external validation evidence to test whether transformer fusion adds clinically meaningful, transportable, and interpretable patient-level information while preserving acceptable discrimination and calibration. The empirical evidence shows a nuanced pattern. A RNA-plus-clinical ridge Cox baseline achieved the strongest apparent discrimination (C-index = 0.724), while the full multi-omics ridge Cox model produced lower aggregate discrimination (C-index = 0.682) but stronger biological interpretability, temporal stability, improved calibration logic, and positive decision-curve value. The transformer models produced moderate internal discrimination (RNA-only C-index = 0.6333; full multi-omics transformer C-index = 0.6094) and modest but non-trivial external transportability on MSK-IMPACT LUAD (C-index = 0.5368), while retaining patient-level mechanistic attribution through Integrated Gradients. Attribution analysis indicated that RNA expression dominated model explanations, contributing approximately 70% of attribution mass, followed by somatic mutations (15%), CNVs (10%), and clinical variables (5%). Subtype-level interpretation was biologically coherent: high-risk cases showed TP53 activity, cell-cycle dysregulation and DNA repair defects; intermediate-risk cases showed metabolic reprogramming; and low-risk cases showed immune-rich signatures. The paper argues that the strongest claim for explainable transformer-based multi-omics LUAD modelling is not simple predictive superiority over classical baselines. Rather, its defensible contribution is a validated, auditable, and patient-level framework that can connect discrimination, calibration, transportability, and biologically intelligible explanation in a form suitable for further prospective evaluation.},
      keywords = {lung adenocarcinoma; multi-omics; transformer; survival prediction; explainable AI; Integrated Gradients; external validation; calibration; precision oncology; TCGA; MSK-IMPACT},
      month = {August},
      doi = {https://doi.org/10.64388/IREV10I2-1722517}
  }