Upstream Probabilistic Meta-Imputation for Multimodal Pediatric Pancreatitis Classification

2025-11-19

AI Summary

Diagnosing pediatric pancreatitis remains challenging due to limited clinical samples and complex, heterogeneous features in multimodal MRI (T1-weighted/T2-weighted). To address data scarcity and model generalizability, we propose an upstream probabilistic meta-completion method: a class-conditional Gaussian mixture model is constructed in a low-dimensional meta-feature space—rather than the raw image space—to enable efficient, high-fidelity synthetic data augmentation. Our lightweight diagnostic pipeline integrates radiomic features, modality-specific logistic regression, and a random forest meta-classifier. Evaluated on 67 pediatric cases, the method achieves an AUC of 0.908 ± 0.072—improving upon the real-data-only baseline by approximately five percentage points. To our knowledge, this is the first work to embed probabilistic generative modeling into the upstream meta-feature space of meta-learning, simultaneously ensuring data efficiency, computational lightness, and clinical interpretability.

Technology Category

• Machine Learning: Calibration & Uncertainty Quantification
• Reasoning under Uncertainty: Relational Probabilistic Models
• Knowledge Representation and Reasoning: Diagnosis and Abductive Reasoning

Application Category

• Semantics and Knowledge: Methods to enhance, augment, integrate or synergize semantic models such as knowledge graphs and LLMs
• User Modeling, Personalization and Recommendation: Explainable and interpretable methods for personalization
• Web Mining and Content Analysis: Large pretrained models with web data

Abstract

Pediatric pancreatitis is a progressive and debilitating inflammatory condition, including acute pancreatitis and chronic pancreatitis, that presents significant clinical diagnostic challenges. Machine learning-based methods also face diagnostic challenges due to limited sample availability and multimodal imaging complexity. To address these challenges, this paper introduces Upstream Probabilistic Meta-Imputation (UPMI), a light-weight augmentation strategy that operates upstream of a meta-learner in a low-dimensional meta-feature space rather than in image space. Modality-specific logistic regressions (T1W and T2W MRI radiomics) produce probability outputs that are transformed into a 7-dimensional meta-feature vector. Class-conditional Gaussian mixture models (GMMs) are then fit within each cross-validation fold to sample synthetic meta-features that, combined with real meta-features, train a Random Forest (RF) meta-classifier. On 67 pediatric subjects with paired T1W/T2W MRIs, UPMI achieves a mean AUC of 0.908 $pm$ 0.072, a $sim$5% relative gain over a real-only baseline (AUC 0.864 $pm$ 0.061).

Problem

• Diagnosing pediatric pancreatitis using multimodal MRI imaging data
• Addressing limited sample availability in machine learning classification
• Handling complexity of multimodal imaging for accurate disease classification

Innovation

• Uses probabilistic meta-imputation in feature space
• Applies Gaussian mixture models for synthetic sampling
• Trains Random Forest classifier with augmented features