Analysis of surgical videos for bilateral axilla–breast approach robotic thyroidectomy using quantum AI/ML
Anzaetek & Seoul National University Bundang Hospital (SNUBH) Quantum Laboratory

Problem

• Endoscopic videos require strict monitoring for real‐time alerts and surgical training.

Challenge

• At top-tier hospitals, anomalies (outliers) are extremely rare, making it difficult for conventional machine learning techniques to learn from such datasets.

Solution

• Developing a quantum hybrid quantum machine learning (QML) approach.

• QML is expected to yield better performance in rare event detection, few-shot learning, and anomaly detection tasks.

Federated Learning (FL)
Federated Learning is a distributed machine learning framework that allows multiple institutions to collaboratively train a model without directly sharing sensitive data. This approach is particularly useful in the healthcare sector, where patient privacy and data security are paramount. For example, when applying FL to predict acute kidney injury (AKI), hospitals and research institutions can combine their unique data insights while complying with data protection regulations. By training on diverse, distributed datasets, FL can build a powerful deep learning (DL) model to identify early risk factors for AKI, ultimately reducing mortality and improving patient outcomes.

Quantum Machine Learning (QML)
Quantum Machine Learning leverages the principles of quantum mechanics to process information more efficiently, offering significant advantages over traditional machine learning (ML) and deep learning (DL). QML algorithms can explore exponentially large solution spaces, providing better results for optimization and pattern recognition tasks compared to conventional methods that require extensive computational time. By integrating QML with federated learning to form Quantum Federated Learning (QFL), it is possible to overcome limitations in distributed learning systems caused by high-dimensional feature spaces or complex model architectures, thereby enhancing performance. QFL is expected to demonstrate excellent suitability, particularly in fields such as AKI prediction, where subtle patterns must be detected across multi-institutional datasets.