arXiv subjects

Trong Nghia Nguyen

Publications and source records attributed to Trong Nghia Nguyen.

4 recordsLinked to original sources

arXiv2026

SMILESGNN: Interpretable Clinical Toxicity Prediction via SMILES-Graph Cross-Attention Fusion

Drug toxicity prediction is critical for reducing late-stage attrition in drug discovery, yet remains challenging due to severe class imbalance, scaffold-based generalization, and the clinical need for interpretable predictions. Single-modality approaches-SMILES Transformers or graph neural networks capture complementary aspects of molecular structure, while sequence-only models cannot directly provide graph-attributed explanations. We present SMILESGNN, a multimodal architecture that fuses a SMILES Transformer encoder and a GATv2 graph encoder via cross-attention, and SMILESGNN-PT, a variant using a ChemBERTa-2 pretrained backbone. The design retains an explicit graph branch within the predictive pipeline, supporting GNNExplainer-based analysis of substructures associated with toxic predictions. On ClinTox, SMILESGNN achieves AUC-ROC 0.987 and F1 0.906 with only 0.4M parameters, performing competitively with a strong SMILESTransformer and a larger ChemBERTa-2/GATv2 concat-fusion baseline. On Tox21 (12 tasks), SMILESGNN-PT obtains mean AUC-ROC 0.750, comparable to ChemBERTa-2 alone and the same-backbone concat-fusion baseline. Overall, the results suggest that cross-attention is a practical fusion alternative that preserves competitive predictive performance while enabling graph-based interpretability support.

cs.LG↗

arXiv2026

A Leakage-Aware Multimodal Evaluation Framework for Early Intraoperative Acute Kidney Injury Prediction

Postoperative acute kidney injury (AKI) after major non-cardiac surgery carries substantial morbidity, yet early intraoperative risk stratification remains difficult. In this retrospective cohort study, we propose SynerT, a waveform-only hybrid temporal backbone that combines a causal dilated TCN with a hierarchy of dilated recurrent layers to encode early intraoperative physiologic trajectories for AKI risk prediction. Building on SynerT, we further design two model variants that extend the backbone with structured clinical context: SynerT-MM, a late-fusion multimodal extension that integrates hemodynamic burden summaries and preoperative covariates, and SynerTStack, a leakage-safe stacked ensemble that combines cross-validated predictions from SynerT-MM with strong tabular baselines at the meta-learning stage. All models are evaluated under a strict leakage-aware framework on VitalDB, a high-fidelity perioperative database, with prediction restricted to information available within the first 60 intraoperative minutes. Among 2,413 waveform-usable cases (180 AKI-positive; 7.46% prevalence), SynerT fell well below strong structured-data baselines, demonstrating that waveform-only temporal modeling is insufficient under strict early constraints. SynerTMM recovered discrimination by incorporating hemodynamic burden summaries and preoperative covariates, and SynerT-Stack achieved the best overall performance across AUROC, AUPRC, and F1-max. Cross-fitted Platt recalibration substantially corrected calibration defects in both multimodal variants, and decision-curve analysis confirmed the recalibrated stacked model delivered the strongest net clinical benefit across low-to-intermediate thresholds.

cs.LG↗

arXiv2026

Faithful Action-unit Causal Reasoning for Counterfactually Faithful Emotion Explanations

Multimodal models can name the action units (AUs) behind a facial emotion, but their AU->emotion rationales are typically plausible rather than faithful: nothing forces the AUs a model invokes to be the AUs that actually drive its prediction. We cast AU->emotion reasoning as a counterfactual-consistency problem between the rationale, the label, and a structural AU->emotion causal graph G, and propose FACR, which grounds the reasoner in an independently induced, polarity-aware G and trains a counterfactual-faithfulness objective: a do-intervention on an AU that G marks causal for a class must move the prediction, while one it marks irrelevant must leave it unchanged. Faithfulness is thereby both trainable and measurable through a matching interventional metric, which we evaluate against a known causal structure, the PSPI pain-AU composition, as no existing affective-reasoning benchmark allows. We are explicit that this metric tests fidelity to the supplied structure rather than its rediscovery: it asks whether the trained reasoner invokes the AUs the structure marks causal, on held-out subjects and a second dataset. Under subject-independent evaluation on UNBC-PAIN, the objective raises the agreement between the invoked AUs and the PSPI composition from a no-objective baseline of 0.08 to 0.57, at a small detection cost; an unfaithfulness control attributes the gain to the objective. On a cross-dataset emotion transfer, the objective likewise raises fidelity to G on a seven-class task (0.50 to 0.84). Finally, we attach a language verbalizer and extend the audit to the generated text: biasing each action unit's emission by its latent activation makes the rationale faithful by construction, so that ablating an AU removes it from the explanation, a property that transfers to a second language-model backbone, whereas a freely generated rationale is unfaithful.

cs.CV↗

arXiv2026

A tissue-informed deep learning-based method for positron range correction in preclinical 68Ga PET imaging

Positron range (PR) limits spatial resolution and quantitative accuracy in PET imaging, particularly for high-energy positron-emitting radionuclides like 68Ga. We propose a deep learning method using 3D residual encoder-decoder convolutional neural networks (3D RED-CNNs), incorporating tissue-dependent anatomical information through a u-map-dependent loss function. Models were trained with realistic simulations and, using initial PET and CT data, generated positron range corrected images. We validated the models in simulations and real acquisitions. Three 3D RED-CNN architectures, Single-channel, Two-channel, and DualEncoder, were trained on simulated PET datasets and evaluated on synthetic and real PET acquisitions from 68Ga-FH and 68Ga-PSMA-617 mouse studies. Performance was compared to a standard Richardson-Lucy-based positron range correction (RL-PRC) method using metrics such as mean absolute error (MAE), structural similarity index (SSIM), contrast recovery (CR), and contrast-to-noise ratio (CNR). CNN-based methods achieved up to 19 percent SSIM improvement and 13 percent MAE reduction compared to RL-PRC. The Two-Channel model achieved the highest CR and CNR, recovering lung activity with 97 percent agreement to ground truth versus 77 percent for RL-PRC. Noise levels remained stable for CNN models (approximately 5.9 percent), while RL-PRC increased noise by 5.8 percent. In preclinical acquisitions, the Two-Channel model achieved the highest CNR across tissues while maintaining the lowest noise level (9.6 percent). Although no ground truth was available for real data, tumor delineation and spillover artifacts improved with the Two-Channel model. These findings highlight the potential of CNN-based PRC to enhance quantitative PET imaging, particularly for 68Ga. Future work will improve model generalization through domain adaptation and hybrid training strategies.

physics.med-ph↗