Quantum Machine Learning and Data Re-Uploading: Evaluation on Benchmark and Laboratory Medicine Data Sets.

Area of Science:

• Quantum Computing
• Machine Learning
• Computational Biology

Background:

• Quantum machine learning (QML) offers potential advantages over classical methods but lacks extensive study with real-world healthcare data.
• This research evaluates the quantum circuits with data re-uploading (QC-REUP) algorithm against classical and other QML approaches.
• Performance was assessed using benchmark and laboratory medicine datasets.

Purpose of the Study:

• To evaluate the performance of the QC-REUP algorithm for classification tasks.
• To compare QC-REUP against classical machine learning (ML) and other QML algorithms.
• To assess the impact of parameter optimization on QC-REUP performance using real-world data.

Main Methods:

• Selected four datasets (2-30 features) for evaluation.
• Conducted baseline classification performance (F1 score) comparisons using QC-REUP, 2 QML, and 4 classical ML algorithms.
• Optimized QC-REUP parameters on a plasma amino acid (PAA) dataset for performance enhancement and final comparison.

Main Results:

• QC-REUP outperformed quantum and linear classical algorithms on low-dimensional data.
• QML performance declined as input dimensionality increased.
• Optimized QC-REUP performed comparably to linear algorithms but was outperformed by nonlinear classical ML algorithms on the PAA dataset.

Conclusions:

• Data re-uploading QML algorithms show comparable performance to classical methods in specific, low-dimensional contexts.
• While optimization aids QC-REUP, significant advancements in quantum hardware and algorithms are necessary for effective application in laboratory medicine and biomedical research.