Speaker
Description
The Q-STARS project brings together companies and research institutions to advance the application of quantum technologies to complex problems in health, energy and industry. Within this consortium, our work centers on two closely related lines of activity, one focused on building the technical infrastructure the project relies on, and the other on applying quantum machine learning to a pressing problem in hospital care.
The first line is the technical leadership of Polypus, an open source library for distributed quantum computing that we are developing as part of Q-STARS. Polypus lets researchers run quantum circuits and train variational algorithms, including VQE, QAOA and QML models, across one or several QPUs, simulated or real, without changing the underlying circuit code. Its core is written in Rust for performance and correctness, while Python bindings keep it accessible to teams already working with Qiskit, so it acts as a drop in accelerator rather than a replacement for existing workflows. Polypus distributes shot execution and population based training automatically across the available quantum processing units, and switches between a local simulator, CESGA's CUNQA platform and CESGA's QMIO real QPU without any change to the algorithm itself. We already have preliminary performance results that support this approach. Batched simulation of full training populations has also shown end to end speedups of between 1.4 and 2.1 times over submitting circuits one by one, with the gain growing as circuits get larger.
The second line is research into applying quantum machine learning to the detection of outbreaks caused by bacteria resistant to multiple antibiotics, a growing threat in hospitals. Working with clinical and microbiological data from real cases, in collaboration with the University Hospital of A Coruña and its research institute, we are exploring how quantum models can identify patients and hospital areas at higher risk of these infections, even when past cases are too few to train reliable classical models. We are studying dimensionality reduction and quantum encoding techniques that make efficient use of the limited qubits available on current hardware, and testing supervised, unsupervised and time series approaches that combine clinical variables such as antibiotic use, invasive procedures and severity scores. The goal is to anticipate outbreaks early enough for hospitals to take preventive measures, and where that is not possible, to help limit their impact.
This work has been supported by the Project Q-STARS (expediente ITC-20251224) funded by CDTI through the Consorcios Regionales INNTERCONECTA-STEP 2025 programme and co-financed by the European Union through the European Regional Development Fund (ERDF) 2021-2027.