Publikation
MiracleNet: A biologically-interpretable machine learning model for resected non-small-cell lung cancer
Rashika Jakhmola; David Antony Selby; Mert Cihan; Dusan Prascevic; Elisabetta Petracci; Paola Ulivi; Enriqueta Felip; Rocío Caro Consuegra; Franco Stella; Piergiorgio Solli; Desideria Argnani; Milena Urbini; Johannes Urban Mayer; Sebastian Vollmer; Christian Martin; Jan Ewald; Maximilian Sprang
In: Computational and Structural Biotechnology Journal, Vol. 0145, No. ja, Pages 1-26, American Association for the Advancement of Science, 6/2026.
Zusammenfassung
Lung cancer remains one of the leading causes of cancer-related mortality worldwide. Ac-curate prediction of relapse is notoriously difficult, posing significant challenges to patientcare and necessitating advanced tools to improve prognostic outcomes. MicroRNA (miRNA) expression profiles hold promise as biomarkers for predicting relapse, yet existing predictive models lack interpretability or sufficient predictive performance. Biologically-informed neural networks have emerged as a modelling approach incorporating biological interpretability and predictive accuracy. Here, we introduce MiracleNet, to our knowledge the first visible neural network in which sparse connectivity is structured by the miRNA→target-gene→pathway hierarchy for disease-free survival prediction from circulating miRNA in non-small-cell lung cancer, and the first to expose interpretable importances jointly at all three biological layers, with nodes connected by prior knowledge about miRNA targets and related biological pathways. Our model, which also integrates clinical data, achieves a maximum concordance index of 0.76 and demonstrates improved generalisation over un-constrained neural networks of the same dimensionality (including both dense and sparse architectures lacking biological knowledge), and provides explicit biological interpretability. Our model also highlights several important biomarkers in the form of predictive miRNAs and connected biological pathways. We additionally evaluate MiracleNet under a nested repeated 80/20 protocol, augmented with patient sex and tumour stage as clinical covariates, and combined across cf- and ev-miRNAs through early and intermediate fusion; these analyses are reported as separate sections.