FAU researchers test quantum methods on lung-cancer data
A preliminary experiment used a small dataset augmented with synthetic samples; clinical validation remains ahead.
Key Takeaways
- •The study began with 309 records and augmented the dataset to 540 samples.
- •The September paper reports experimental results, not clinical diagnostic accuracy.
- •FAU discussed the research in a Nov. 2 announcement.
Florida Atlantic University researchers have tested quantum computing methods against a small lung-cancer dataset, exploring how models perform when training data are scarce. The university described the preliminary work on Nov. 2.
The peer-reviewed paper, published Sept. 9 in Discover Quantum Science, started with 309 patient records containing symptoms and risk factors. Researchers balanced and expanded the dataset to 540 samples using synthetic data. Those additional samples are not newly recruited patients.
The team compared three classifiers: quantum support vector machines, quantum k-nearest neighbors and quantum boosting. The paper describes five rounds of evaluation in which each model trained on 108 samples and was tested on the remaining 432, rotating the training portion across the dataset.
The authors reported their strongest support-vector-machine result at 94.31% accuracy in simulation and 91.90% using D-Wave hardware. The numbers reflect this experimental evaluation, not diagnostic accuracy in clinical practice. FAU says future work should assess independent clinical datasets and how to make the models’ reasoning more interpretable.
FAU professor Arslan Munir co-authored the study with Muhammad Minoar Hossain and Md. Hasibul Hassan Himal.
