arXiv subjects

Taisei Tosaki

Publications and source records attributed to Taisei Tosaki.

3 recordsLinked to original sources

arXiv2026

Supervised Graph Contrastive Learning for Gene Regulatory Networks

Graph Contrastive Learning (GCL) is a powerful self-supervised learning framework that performs data augmentation through graph perturbations, with growing applications in the analysis of biological networks such as Gene Regulatory Networks (GRNs). The artificial perturbations commonly used in GCL, such as node dropping, induce structural changes that can diverge from biological reality. This concern has contributed to a broader trend in graph representation learning toward augmentation-free methods, which view such structural changes as problematic and should be avoided. However, this trend overlooks the fundamental insight that structural changes from biologically meaningful perturbations are not a problem to be avoided, but rather a rich source of information, thereby ignoring the valuable opportunity to leverage data from real biological experiments. Motivated by this insight, we propose SupGCL (Supervised Graph Contrastive Learning), a new GCL method for GRNs that directly incorporates biological perturbations from gene knockdown experiments as supervision. SupGCL is a probabilistic formulation that continuously generalizes conventional GCL, linking artificial augmentations with real perturbations measured in knockdown experiments, and using the latter as explicit supervision. On patient-derived GRNs from three cancer types, we train GRN representations with SupGCL and evaluate it in two regimes: (i) embedding space analysis, where it yields clearer disease-subtype structure and improves clustering, and (ii) task-specific fine-tuning, where it consistently outperforms strong graph representation learning baselines on 13 downstream tasks spanning gene-level functional annotation and patient-level prediction.

cs.LG↗

arXiv2026

A Nationwide Japanese Medical Claims Foundation Model: Balancing Model Scaling and Task-Specific Computational Efficiency

Clinical risk prediction using longitudinal medical data supports individualized care. Self-supervised foundation models have emerged as a promising approach for leveraging large-scale unlabeled healthcare records. In natural language processing, scaling laws suggest that larger models achieve predictably lower pretraining losses, supporting the foundation model paradigm. However, for structured medical data, characterized by a limited vocabulary and sparse observations, whether increasing model size consistently improves downstream predictions is unclear, as most studies evaluate only a single model scale. In this study, we evaluated the relationship between model scale and downstream task performance for structured medical foundation models. Using a random sample (2.3 million patients, 32 hospitals) from a nationwide 519-hospital Japanese claims database, we pretrained encoder-only Transformers at five scales (2.2M-101M parameters) for disease incidence and medication prediction. Downstream performance saturated at task-dependent thresholds: disease prediction benefited from larger models (32M-101M), whereas medication prediction saturated at 11M, reducing pretraining time by 178 h. Across all tasks, the best-performing model consistently outperformed a Light Gradient Boosting Machine baseline in the area under the precision-recall curve. These findings indicate that, unlike the monotonically decreasing pretraining loss, the optimal model size varied depending on task characteristics. This task-dependent saturation provides practical guidance for balancing predictive performance and computational cost in structured medical foundation models.

cs.LG↗

arXiv2025

Out-of-distribution Reject Option Method for Dataset Shift Problem in Early Disease Onset Prediction

Machine learning is increasingly used to predict lifestyle-related disease onset using health and medical data. However, its predictive accuracy for use is often hindered by dataset shift, which refers to discrepancies in data distribution between the training and testing datasets. This issue leads to the misclassification of out-of-distribution (OOD) data. To diminish dataset shift in real-world settings, this paper proposes the out-of-distribution reject option for prediction (ODROP). This method integrates an OOD detection model to preclude OOD data from the prediction phase. We used two real-world health checkup datasets (Hirosaki and Wakayama) with dataset shift, across three disease onset prediction tasks: diabetes, dyslipidemia, and hypertension. Both components of ODROP method -- the OOD detection model and the prediction model -- were trained on the Hirosaki dataset. We assessed the effectiveness of ODROP on the Wakayama dataset using AUROC-rejection rate curve plot. In the five OOD detection approaches (the variational autoencoder, neural network ensemble std, neural network ensemble epistemic, neural network energy, and neural network gaussian mixture based energy measurement), the variational autoencoder method demonstrated notably higher stability and a greater improvement in AUROC. For example, in the Wakayama dataset, the AUROC for diabetes onset increased from 0.80 without ODROP to 0.90 at a 31.1% rejection rate, and for dyslipidemia, it improved from 0.70 without ODROP to 0.76 at a 34% rejection rate. In addition, we categorized dataset shifts into two types using SHAP clustering -- those that considerably affect predictions and those that do not. This study is the first to apply OOD detection to actual health and medical data, demonstrating its potential to substantially improve the accuracy and reliability of disease prediction models amidst dataset shift.

cs.LG↗