Canonical residual physics-informed neural networks and algorithmic perturbation landscapes map cell-fate transitions in longitudinal H3K27M glioma
Abstract
Pediatric high-grade glioma is a dynamically heterogeneous disease in which therapeutic selection pressures, microenvironmental selection and developmental plasticity can reorganize cellular-state structure over time. Here we investigated these adaptive behaviors in glioma dynamics from single-cell ecologies by merging physics-driven artificial intelligence and complex systems theory. Specifically, this study explores canonical residual-based physics-informed neural networks (PINNs), dynamical systems modelling comparators, explainable artificial intelligence (XAI), and Block Decomposition Method (BDM) perturbation analysis to understand glioma cell fate decisions. The primary analyses distinguished longitudinal initial from recurrent H3K27M cells and, within the recurrent specimen set, stem/progenitor-like from differentiated/glial-like annotation-derived states. Across five-fold cross-validation, RD achieved AUC 0.995 ± 0.005 and macro-F1 0.973 ± 0.016 for initial--recurrence discrimination, and AUC 0.698 ± 0.012 and macro-F1 0.629 ± 0.014 for recurrent-state subtype discrimination. FP achieved AUC 0.931 ± 0.018 and 0.665 ± 0.012, respectively. Cross-model XAI repeatedly prioritized PTGDS, CHI3L1, IGFBP7, SPARCL1 and NRXN3 in longitudinal progression. BDM perturbation rankings emphasized heat-shock/stress-response and immune-associated network signatures. These findings are candidate computational network biomarkers rather than experimentally established causal signatures. The study provides a reproducible bridge between physics-constrained latent dynamics, interpretable gene attribution, and data-driven algorithmic perturbation landscapes for in silico pattern generation about glioma cell fate plasticity and early recurrence detection.
Related
- Changes in physical activity and transitions between functional states among adults aged 50 years or older: a longitudinal study of five aging cohorts
- Nurse-Led Longitudinal Screening and Care Transitions for Post-Traumatic Stress Symptoms After Acute Type A Aortic Dissection: A 12-Month Retrospective Longitudinal Cohort Study
- The dynamics of loneliness: A longitudinal exploration of transitions in Japan across the COVID-19 period, 2020–2024
- Metabolic and functional correlates of first transitions to low muscle strength and memory reserve across four longitudinal cohorts
- Multidimensional health latent transitions and their influencing factors among middle-aged and older adults with chronic diseases from the perspective of healthy aging: a longitudinal analysis of CHARLS data
BibTeX
@misc{uthamacumaran2026canonical,
title = {{Canonical residual physics-informed neural networks and algorithmic perturbation landscapes map cell-fate transitions in longitudinal H3K27M glioma}},
author = {Abicumaran Uthamacumaran},
howpublished = {arXiv},
year = {2026},
month = oct,
doi = {10.48550/arxiv.2610.33000},
eprint = {2610.33000},
archivePrefix = {arXiv},
url = {https://arxiv.org/abs/2610.33000},
}
RIS
TY - UNPB
TI - Canonical residual physics-informed neural networks and algorithmic perturbation landscapes map cell-fate transitions in longitudinal H3K27M glioma
AU - Abicumaran Uthamacumaran
PY - 2026
DA - 2026/10/06
PB - arXiv
DO - 10.48550/arxiv.2610.33000
UR - https://arxiv.org/abs/2610.33000
AB - Pediatric high-grade glioma is a dynamically heterogeneous disease in which therapeutic selection pressures, microenvironmental selection and developmental plasticity can reorganize cellular-state structure over time. Here we investigated these adaptive behaviors in glioma dynamics from single-cell ecologies by merging physics-driven artificial intelligence and complex systems theory. Specifically, this study explores canonical residual-based physics-informed neural networks (PINNs), dynamical systems modelling comparators, explainable artificial intelligence (XAI), and Block Decomposition Method (BDM) perturbation analysis to understand glioma cell fate decisions. The primary analyses distinguished longitudinal initial from recurrent H3K27M cells and, within the recurrent specimen set, stem/progenitor-like from differentiated/glial-like annotation-derived states. Across five-fold cross-validation, RD achieved AUC 0.995 ± 0.005 and macro-F1 0.973 ± 0.016 for initial--recurrence discrimination, and AUC 0.698 ± 0.012 and macro-F1 0.629 ± 0.014 for recurrent-state subtype discrimination. FP achieved AUC 0.931 ± 0.018 and 0.665 ± 0.012, respectively. Cross-model XAI repeatedly prioritized PTGDS, CHI3L1, IGFBP7, SPARCL1 and NRXN3 in longitudinal progression. BDM perturbation rankings emphasized heat-shock/stress-response and immune-associated network signatures. These findings are candidate computational network biomarkers rather than experimentally established causal signatures. The study provides a reproducible bridge between physics-constrained latent dynamics, interpretable gene attribution, and data-driven algorithmic perturbation landscapes for in silico pattern generation about glioma cell fate plasticity and early recurrence detection.
ER -