Predicting pesticide environmental fate properties—specifically soil degradation halflife (DegT50) and organic carbon sorption coefficient (Koc)—is essential for regulatory risk assessment under EC 1107/2009. Classical QSAR models achieve moderate predictive power for Koc but struggle with DegT50 due to its dependence on complex soil microbial processes. This work investigates whether variational quantum circuits (VQCs) can capture non-linear molecular descriptor relationships that elude classical regressors. We encode 21 physicochemical descriptors for 110 EU-registered pesticides into per-target quantum circuits: an 8-qubit circuit for DegT50 (17 core features) and a 12-qubit circuit for Koc (all 21 features), each with IQP-style encoding and data re-uploading. Leave-one-out crossvalidation benchmarks reveal that Random Forest (RF) achieves R2 = 0.287 for DegT50 and 0.712 for Koc, while Gradient Boosting (GB) reaches 0.283 and 0.750, respectively (fair 20-feature set, excluding circular bioaccessibility). An MLP neural network achieves only 0.075 and 0.497. The VQC alone underperforms all classical baselines (R2 = −0.028 for DegT50, 5-fold CV). A hybrid stacking framework with nested LOO-CV for leakage free α optimization yields α = 0.20±0.02, but the resulting hybrid R2 (0.288 for DegT50) improves over standalone RF by only +0.001—a negligible and statistically insignificant margin. These results demonstrate that near-term VQCs on small regulatory datasets (N ≈ 110) cannot outperform properly feature-engineered classical models, and caution against premature claims of quantum advantage in chemistry applications with limited training data.
Predicting pesticide environmental fate properties—specifically soil degradation halflife (DegT50) and organic carbon sorption coefficient (Koc)—is essential for regulatory risk assessment under EC 1107/2009. Classical QSAR models achieve moderate predictive power for Koc but struggle with DegT50 due to its dependence on complex soil microbial processes. This work investigates whether variational quantum circuits (VQCs) can capture non-linear molecular descriptor relationships that elude classical regressors. We encode 21 physicochemical descriptors for 110 EU-registered pesticides into per-target quantum circuits: an 8-qubit circuit for DegT50 (17 core features) and a 12-qubit circuit for Koc (all 21 features), each with IQP-style encoding and data re-uploading. Leave-one-out crossvalidation benchmarks reveal that Random Forest (RF) achieves R2 = 0.287 for DegT50 and 0.712 for Koc, while Gradient Boosting (GB) reaches 0.283 and 0.750, respectively (fair 20-feature set, excluding circular bioaccessibility). An MLP neural network achieves only 0.075 and 0.497. The VQC alone underperforms all classical baselines (R2 = −0.028 for DegT50, 5-fold CV). A hybrid stacking framework with nested LOO-CV for leakage free α optimization yields α = 0.20±0.02, but the resulting hybrid R2 (0.288 for DegT50) improves over standalone RF by only +0.001—a negligible and statistically insignificant margin. These results demonstrate that near-term VQCs on small regulatory datasets (N ≈ 110) cannot outperform properly feature-engineered classical models, and caution against premature claims of quantum advantage in chemistry applications with limited training data.
Predicting pesticide environmental fate properties—specifically soil degradation halflife (DegT50) and organic carbon sorption coefficient (Koc)—is essential for regulatory risk assessment under EC 1107/2009. Classical QSAR models achieve moderate predictive power for Koc but struggle with DegT50 due to its dependence on complex soil microbial processes. This work investigates whether variational quantum circuits (VQCs) can capture non-linear molecular descriptor relationships that elude classical regressors. We encode 21 physicochemical descriptors for 110 EU-registered pesticides into per-target quantum circuits: an 8-qubit circuit for DegT50 (17 core features) and a 12-qubit circuit for Koc (all 21 features), each with IQP-style encoding and data re-uploading. Leave-one-out crossvalidation benchmarks reveal that Random Forest (RF) achieves R2 = 0.287 for DegT50 and 0.759 for Koc, while Gradient Boosting (GB) reaches 0.283 and 0.752, respectively. An MLP neural network achieves only 0.075 and 0.497. The VQC alone underperforms all classical baselines (R2 = −0.028 for DegT50, 5-fold CV). A hybrid stacking framework with nested LOO-CV for leakage-free α optimization yields α = 0.20±0.02, but the resulting hybrid R2 (0.288 for DegT50) improves over standalone RF by only +0.001—a negligible and statistically insignificant margin. These results demonstrate that near-term VQCs on small regulatory datasets (N ≈ 110) cannot outperform properly feature-engineered classical models, and caution against premature claims of quantum advantage in chemistry applications with limited training data.
Predicting pesticide environmental fate properties specifically soil degradation half-life (DegT50) and organic carbon sorption coefficient (K oc)-is essential for regulatory risk assessment under EC 1107/2009. Classical QSAR models achieve moderate predictive power for K oc but struggle with DegT50 due to its dependence on complex soil microbial processes. This work investigates whether variational quantum circuits (VQCs) can capture non-linear molecular descriptor relationships that elude classical regressors. We encode 17 physicochemical descriptors for 111 EU-registered pesticides into a 12-qubit IQP-style quantum circuit with data re-uploading and 8 variational layers (301 trainable parameters). Leave-one-out cross-validation benchmarks reveal that Random Forest (RF) achieves R 2 = 0.194 for DegT50 and 0.726 for K oc (on a circularity-free 16-feature set excluding the bioaccessibility proxy), while Gradient Boosting (GB) reaches 0.223 and 0.764, respectively. The VQC underperforms classical baselines on DegT50 (R 2 < 0, 5-fold CV), consistent with severe overfitting expected from the unfavorable parameter-to-sample ratio (N params /N data ≈ 2.7). A hybrid stacking framework blending QML and RF predictions yields α = 0.20, but the resulting R 2 improvement (+0.02) is not statistically significant. We also assess adverse outcome pathway (AOP) features derived from AOP-Wiki data as additional descriptors, finding no improvement over structural features alone. These results provide a rigorous benchmark for near-term quantum machine learning applied to regulatory chemistry and highlight the critical role of dataset size in VQC generalization.
Predicting drug-target binding affinity with limited training data remains a central challenge in computational drug discovery. We introduce a hybrid quantum-classical framework combining molecular descriptors with variational quantum circuits for interpretable binding affinity prediction. Seven physicochemical descriptors (MW, logP, TPSA, HBD/HBA, rotatable bonds, aromatic rings) were encoded into 6-qubit variational circuits using parameterized rotations and controlled-Z entanglement. Quantum kernels were extracted and used for regression on 1,200 BindingDB molecules across kinase, GPCR, and protease targets, with external validation on 300 ChEMBL compounds. Performance was compared against SVR, random forest, gradient boosting, and neural network baselines using 100-fold bootstrap validation. Variational quantum regression achieved MSE = 0.056 ± 0.009, outperforming SVR by 32% (0.082 ± 0.013), RF by 28%, and GBM by 24%. Quantum advantage was most significant in low-data regimes (n < 500), where VQR maintained R2 > 0.85 versus R2 < 0.72 for classical methods. External validation confirmed generalization (R2 = 0.83). Our Explainable Quantum Pharmacology framework revealed TPSA and logP as dominant predictive features through gradientbased sensitivity analysis, aligning with established medicinal chemistry principles. Quantum kernels provide measurable, chemically interpretable improvements for molecular property prediction in data-limited scenarios. This work demonstrates practical quantum advantage for early-stage drug discovery applications on near-term quantum hardware.
Recent advances in quantum machine learning (QML) offer new opportunities for chemical property prediction, yet their practical performance relative to established classical models remains poorly characterized on real-world datasets. Building on our previous benchmark of classical and deep learning models for collision cross-section (CCS) predictions, we present an empirical comparison of quantum kernel regression (QKR) and a variational quantum regressor (VQR) against classical machine learning (CML) techniques, including Random Forest (RF), support vector regression (SVR), Lasso, and a Voting Regressor (VR) ensemble. Using an aquatic dissolved organic matter (DOM) dataset with five molecular descriptors and five chemical classes, we systematically investigate QKR feature map repetitions and classical readout kernels, as well as VQR circuit depths and optimizer choices, to identify performance-optimal QML configurations. Regression performance is evaluated using the mean absolute error (MAE), root mean square error (RMSE), and coefficient of determination (R2) across training-set sizes up to 600 molecules per class. Overall, RF and Lasso deliver the most robust CCS predictions, achieving R2 ≥ 0.9 for lignins, carbohydrates, lipids, and proteins. Meanwhile, the optimized QKR architecture exhibits substantial gains as data increase and achieves competitive performance on chemically homogeneous subsets such as carbohydrates and lipids. In contrast, the VQR ansatz remains consistently weaker than both QKR and the classical baselines across all classes and data sizes. The results highlight the strong influence of dataset size, feature representation, and molecular heterogeneity on QML performance while also identifying conditions under which quantum methods may become competitive in future studies.
Accurate prediction of bond dissociation energies (BDEs) underpins mechanistic insight and the rational design of molecules and materials. We present a systematic, reproducible benchmark comparing quantum and classical machine learning models for BDE prediction using a chemically curated feature set encompassing atomic properties (atomic numbers, hybridization), bond characteristics (bond order, type), and local environmental descriptors. Our quantum framework, implemented in Qiskit Aer on six qubits, employs ZZFeatureMap encodings with variational ansätz (RealAmplitudes) across multiple architectures Variational Quantum Regressors (VQR), Quantum Support Vector Regressors (QSVR), Quantum Neural Networks (QNN), Quantum Convolutional Neural Networks (QCNN), and Quantum Random Forests (QRF). These are rigorously benchmarked against strong classical baselines, including Support Vector Regression (SVR), Random Forests (RF), and Multi-Layer Perceptrons (MLP). Comprehensive evaluation spanning absolute and relative error metrics, threshold accuracies, and error distributions shows that top-performing quantum models (QCNN, QRF) match the predictive accuracy and robustness of classical ensembles and deep networks, particularly within the chemically prevalent mid-range BDE regime. These findings establish a transparent baseline for quantum-enhanced molecular property prediction and outline a practical foundation for advancing quantum computational chemistry toward near-chemical accuracy.
This research presents a hybrid variational quantum machine learning framework for real-world soil condition prediction using hyperspectral and environmental data. The study integrates classical preprocessing techniques with a Variational Quantum Classifier (VQC) implemented using Qiskit to classify soil moisture conditions into dry and wet states. A real soil moisture dataset containing environmental and spectral features was utilized to evaluate the model’s performance. Data normalization and feature encoding were applied before embedding the features into a parameterized quantum circuit composed of a ZZFeatureMap and TwoLocal ansatz. Model optimization was performed using the COBYLA optimizer within a simulator-based quantum backend. The system demonstrates the feasibility of applying near-term quantum machine learning techniques to agricultural analytics. Performance evaluation includes confusion matrix analysis, accuracy, precision, recall, and F1-score metrics. The study further discusses scalability considerations, simulator-based execution constraints, and the implications of deploying the approach on real quantum hardware. This work contributes to the intersection of quantum computing, precision agriculture, and hyperspectral data analytics, highlighting the potential of hybrid quantum-classical workflows for next-generation intelligent farming systems.