Quantum Kernel Advantage over Classical Collapse in Medical Foundation Model Embeddings

2026-04-27

AI Summary

This study addresses the performance collapse of classical linear kernels in binary medical image classification under severe class imbalance. To enable a fair comparison, the authors propose a two-stage evaluation framework that freezes embeddings from a medical foundation model (e.g., MedSigLIP-448) and applies identical PCA dimensionality reduction before systematically comparing quantum support vector machines (QSVMs) against classical SVMs for the first time. Under zero hyperparameter tuning, QSVMs consistently outperform linear SVMs across all 18 experimental configurations, achieving an average F1 score of 0.343 versus 0.050 while effectively avoiding class collapse. This advantage remains robust under simulated noise and is further corroborated by kernel effective rank analysis and statistically significant hypothesis testing.

Technology Category

• Machine Learning: Quantum Machine Learning
• Computer Vision: Medical and Biological Imaging
• Data Mining & Knowledge Management: Data Compression

Application Category

• Search and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for ranking
• Web Mining and Content Analysis: Robustness and generalizability of Web mining methods
• Economics, Online Markets and Human Computation: Data quality aspects of human-annotated datasets

Abstract

We provide evidence of quantum kernel advantage under noiseless simulation in binary insurance classification on MIMIC-CXR chest radiographs using quantum support vector machines (QSVM) with frozen embeddings from three medical foundation models (MedSigLIP-448, RAD-DINO, ViT-patch32). We propose a two-tier fair comparison framework in which both classifiers receive identical PCA-q features. At Tier 1 (untuned QSVM vs. untuned linear SVM, C = 1 both sides), QSVM wins minority-class F1 in all 18 tested configurations (17 at p < 0.001, 1 at p < 0.01). The classical linear kernel collapses to majority-class prediction on 90-100% of seeds at every qubit count, while QSVM maintains non-trivial recall. At q = 11 (MedSigLIP-448 plateau center), QSVM achieves mean F1 = 0.343 vs. classical F1 = 0.050 (F1 gain = +0.293, p < 0.001) without hyperparameter tuning. Under Tier 2 (untuned QSVM vs. C-tuned RBF SVM), QSVM wins all seven tested configurations (mean gain +0.068, max +0.112). Eigenspectrum analysis reveals quantum kernel effective rank reaches 69.80 at q = 11, far exceeding linear kernel rank, while classical collapse remains C-invariant. A full qubit sweep reveals architecture-dependent concentration onset across models. Code: https://github.com/sebasmos/qml-medimage

Problem

• quantum kernel advantage
• classical collapse
• medical foundation models
• imbalanced classification
• support vector machines

Innovation

• quantum kernel advantage
• medical foundation models
• quantum support vector machine
• classical collapse
• fair comparison framework