Quantum model identifies lung cancer risk profiles with 91.9% accuracy on a physical quantum computer
Researchers at Florida Atlantic University built a quantum machine learning model that flags lung cancer risk profiles from symptom records, reaching 91.90% accuracy on a physical D-Wave quantum computer.
Newsnibs Staff
- 2026-11-17
- Updated 1:02 PM UTC
Researchers at Florida Atlantic University in the United States have built a quantum machine learning approach that identifies lung cancer risk profiles from a small dataset. The best-performing method, a Quantum Support Vector Machine (QSVM) run on D-Wave's quantum annealing technology, made correct predictions 94.31% of the time in simulation and 91.90% on a physical quantum computer. The study was published in the journal Discover Quantum Science.
The team started with a database of 309 medical records listing symptoms such as cough, fatigue and chest pain. They generated synthetic data to balance the risk profiles, which brought the dataset to 540 examples. The models were trained on 108 of those examples, and the rest of the data was used for evaluation through cross-validation.
The study compared three quantum machine learning methods across five computing environments. QSVM on D-Wave's annealing hardware gave the highest accuracy of the combinations tested.
The dataset used only symptoms and risk factors. It left out lung scans, biological analyses and biopsy results, and part of it was synthetic. The authors say the results should be interpreted with caution.
The team plans to test the methods on larger, independent medical databases. In later research they intend to add imaging and genomic data.
Separately, a list of 13 ways self-tuning quantum computers could change business was published on Monday by Dr. Pravir Malik, founder of QIQuantum. It draws on input from members of a quantum computing group that includes executives from Walmart, Morgan Stanley, Citigroup and Akamai Technologies. Self-tuning, in this context, means agentic AI calibrates quantum hardware on its own, which reduces the need for specialized human intervention.
The contributors say removing calibration downtime would let tasks such as dynamic supply chain logistics and real-time risk modeling run continuously. They also say relying less on scarce quantum physicists would lower operating costs and make the technology more commercially viable for enterprises. Participants added that better hardware utilization would send more budget to running experiments and less to manual maintenance.
The use cases they mention include retail inventory optimization, satellite scheduling for space companies and drug discovery. In drug discovery, real-time recalibration would keep a machine from going offline during peak computational demand. The stated goal is to move quantum computing from an experimental science project to a dependable, economically viable enterprise utility.
Photo by Ilya Pavlov on Unsplash