Anastomotic leak is a life-threatening complication following colorectal surgery. This study benchmarks Quantum Neural Networks (QNNs) against hyperparameter-tuned classical models (logistic regression, multi-layer perceptrons, boosting algorithms) for anastomotic leak prediction. Using a 200-patient clinical dataset strictly bounded by a priori medical constraints, we simulated QNNs with ZZFeatureMap encoding and EfficientSU2/RealAmplitudes ansatze under realistic hardware noise. To ensure statistical reliability, performance metrics were averaged across 10 independent optimization runs. The EfficientSU2-BFGS configuration achieved the highest mean AUC of $$0.797 \pm 0.024$$ , while RealAmplitudes with CMA-ES maximized Average Precision ( $$0.504 \pm 0.121$$ ). Crucially, at a fixed, clinically necessary sensitivity of $$83\%$$ , specific QNN configurations achieved significantly higher specificity (up to $$66\%$$ ) and Negative Predictive Value (up to $$96\%$$ ) compared to classical models (maximum $$44\%$$ and $$94\%$$ , respectively), effectively minimizing false positives. However, classical models maintained superior probability calibration for continuous risk stratification. We conclude that QNNs offer robust discriminative performance for clinical screening, warranting further validation on larger, independent cohorts.
This work deals with the design of a hybrid classification model that uses two complementary parallel data processing branches. The aim was to verify whether the connection of different input representations within a common decision mechanism can support the stability and reliability of classification. The outputs of both branches are continuously integrated and together form the final decision of the model. On the validation set, the model achieved accuracy 0.9750, precision 1.0000, recall 0.9500 and F1-score 0.9744 at a threshold value of 0.5. These results suggest that parallel, complementary processing may be a promising direction for further development and optimization of the model, especially in tasks requiring high accuracy while maintaining robust detection of positive cases.
The optimization of Variational Quantum Eigensolver is severely challenged by finite-shot sampling noise, which distorts the cost landscape, creates false variational minima, and induces statistical bias called winner's curse. We investigate this phenomenon by benchmarking eight classical optimizers spanning gradient-based, gradient-free, and metaheuristic methods on quantum chemistry Hamiltonians H$_2$, H$_4$ chain, LiH (in both full and active spaces) using the truncated Variational Hamiltonian Ansatz. We analyze difficulties of gradient-based methods (e.g., SLSQP, BFGS) in noisy regimes, where they diverge or stagnate. We show that the bias of estimator can be corrected by tracking the \textit{population mean}, rather than the biased best individual when using population based optimizer. Our findings, which are shown to generalize to hardware-efficient circuits and condensed matter models, identify adaptive metaheuristics (specifically CMA-ES and iL-SHADE) as the most effective and resilient strategies. We conclude by presenting a set of practical guidelines for reliable VQE optimization under noise, centering on the co-design of physically motivated ansatz and the use of adaptive optimizers.
This paper provides a structural and methodological audit of several quantum-inspired Differential Evolution variants, focusing primarily on the recently proposed PSEQADE algorithm. Our analysis shows that these frameworks share a foundational core characterized by mathematical contradictions and procedural ambi- guities. Specifically, the published text introduces unindexed circular dependencies, dimensional conflations between scalars and vectors, a logically inverted sampling operator, initialization biases, and a monotonically increasing scale factor that causes immediate floating-point overflow under standard execution. Method- ologically, the empirical validation deviates from standard IEEE CEC protocols by evaluating the algorithm on simplified, unshifted base functions while subjecting established baseline optimizers to a limited function evaluation budget. Compounded by the absence of replication code, these discrepancies make independent reproduction unresolvable. Additionally, the reported results, achieving absolute zero variance and perfec- tion on complex, high-dimensional landscapes within a fraction of standard evaluation budgets, are unusual and not discussed in much detail. This analysis indicates that the current formulations do not align with the rigorous validation standards established by the global swarm and evolutionary optimization commu- nities. Furthermore, the manuscript relies entirely on classical, two-dimensional trigonometric operations while asserting unsupported claims of physical quantum advantage. This mischaracterization transforms a standard quantum-inspired conceptual metaphor into a misleading claim of literal quantum utility, creating a fundamental contradiction with the algorithm’s actual classical mechanics.