Quantum machine learning (QML) is an emerging field combining quantum computing and artificial intelligence, with promising applications in medicine and healthcare. This survey reviews more than 60 studies published between 2018 and 2025, highlighting a sharp increase in research activity, especially in the last three years. We address seven core research questions related to publication trends, the use of real quantum hardware versus simulators, quantum architectures overview, dataset types, medical domains, algorithmic frameworks, and reported results. Our analysis shows that most QML research in healthcare is conducted on simulators due to limited hardware access, and it relies on small datasets. Quantum convolutional neural network (QCNN) architectures dominate image-based medical tasks such as tumor detection, pneumonia diagnosis, and ECG interpretation, while feature-based datasets are mainly analyzed with variational quantum classifiers and quantum support vector machines. Despite hardware constraints, QML models often match or surpass classical machine learning approaches in accuracy, frequently reaching 95–99%. However, these performance statements should be qualified to recognize experimental limitations and avoid excessive optimism and should not be interpreted as definitive proof of quantum superiority at this stage. Additionally, issues with reproducibility and reporting of hardware details persist, which is a significant research gap. This review emphasizes the need for standardized benchmarks, more real hardware testing, and architecture-aware algorithm design. With the potential for accelerated diagnostics and personalized healthcare, QML represents a strategic direction for future medical research.
With current technological advancements, almost all new buildings require periodic maintenance inspections of their installed facilities. These inspections are performed by technicians with varying experience, and therefore changing work hours and rates. However, such inspections often cannot be performed while the building is in use, which requires scheduling visits based on available time windows. In this paper, the described vehicle routing problem with time windows and varying employee rates is formulated using the MILP model. Four solution methods are proposed: a mathematical solver, an adapted Tabu Search (TS) algorithm, an adapted 1nn algorithm, and the Artificial Bee Colony (ABC) algorithm. The purpose of the study is to examine the impact of time window lengths and the junior-senior ratio of the technician groups on the final solution. The results show that for larger instances of the problem, it is not possible to use Gurobi as a solving method due to long computational time. Two implemented metaheuristics, Tabu Search and Artificial Bee Colony algorithm, are comparable with the note that Tabu Search usually achieves better results, depending on the specific spatial characteristics of the instance. The best results were obtained with InsertMove operator for neighborhood exploration for both metaheuristics. In our case, it is proved that the longer tabu list size is applied, the better are the results. Tuning the ABC algorithm showed that it performs best with a number of bees of 75 and a larger limit parameter. Future research may include hybrid use of the proposed metaheuristics and multi-objective optimization to investigate the influence of individual objective function coefficients.
This paper presents an experimental evaluation of custom quantum machine learning solutions for binary classification on limited-qubit quantum hardware. To avoid the additional effects introduced by feature selection and dimensionality reduction, nine synthetic datasets were employed, spanning three levels of classification difficulty (easy, medium, and complex) and three feature dimension-alities (3, 4, and 5). This design enabled direct feature-to-qubit encoding and a systematic investigation of how problem complexity affects classification performance. The primary objective was to evaluate the developed solutions on Odra 5, a five-qubit gate-based superconducting quantum processor. Two commonly used variational circuit architectures, RealAmplitudes and EfficientSU2, were compared with Spider, a custom topology-tailored ansatz designed to match the processor’s connectivity. The evaluationexamined classification accuracy, training dynamics, and the influence of the number of variational layers. Experiments were conducted using both FakeOdra, a noise-aware simulator calibrated using measurements from the physical processor, and the Odra5 QPU. Additionally, the temporal stability of classification results was investigated, and exploratory experiments involving direct model training on the QPU were performed. For the easy three-feature dataset, the Spider architecture achieved a validation accuracy of 93.10 ± 5.52% on the physical processor with five variational layers, substantially outperforming the other evaluated architectures. These findings highlight the potential benefits of hardware-aware variational circuit design and emphasize the importance of validating QML models on physical quantum devices.