Quantum Neural Network

QNN performance was high across feature sets, with the combined Sample 3 feature set generally performing best. The study frames QNNs as parameter-efficient compared with classical neural networks and random forests in Sample 3. Sample 3 QNN models achieved test accuracy between 0.9636 and 0.9697 and test AUC between 0.9903 and 0.9960. The study presents the QNN work as simulation and NISQ-oriented rather than deployment on mature fault-tolerant quantum hardware. The QNN used amplitude encoding, parameterized rotations, CNOT entanglement, and Pauli-Z measurement followed by a classical sigmoid layer. The study tested whether a hybrid quantum-classical neural network could classify LUAD versus LUSC using compact biologically meaningful features.