The SH-SY5Y neuroblastoma cell line is often used as an in vitro model of neuronal function and is widely applied to study the molecular events leading to Alzheimer’s disease (AD). Indeed, recently, basic research on SH-SY5Y cells has provided interesting insights for the discovery of new drugs and biomarkers for improved AD treatment and diagnosis. At the same time, untargeted NMR metabolomics is widely applied to metabolic profile analysis and screening for differential metabolites, to discover new biomarkers. In this paper, a compression technique based on convolutional autoencoders is proposed, which can perform a high dimensionality reduction in the spectral signal (up to more than 300 times), maintaining informative features (guaranteed by a reconstruction error always smaller than 5%). Moreover, before compression, an ad hoc preprocessing method was devised to remedy the scarcity of available data. The compressed spectral data were then used to train some SVM classifiers to distinguish diseased from healthy cells, achieving an accuracy close to 78%, a significantly better performance with respect to using standard PCA-compressed data.

Alzheimer’s disease (AD) is a multifactorial neurodegenerative disorder characterized by metabolic dysregulation, oxidative stress, amyloid-β (Aβ) aggregation, metal dyshomeostasis, and mitochondrial dysfunction. Current treatments provide only symptomatic relief, highlighting the need for novel therapeutic strategies. This study investigates the metabolic effects of the alkaloids galantamine (GAL) and lycorine (LYC) in differentiated SH-SY5Y neuroblastoma cells, an established in vitro model for AD, which acquire a neuronal phenotype upon differentiation. Using untargeted and targeted NMR-based metabolomics combined with multivariate statistical analysis, we analyzed extracellular metabolic profiles under basal conditions and following Aβ42 exposure, both in the presence and absence of GAL and LYC. Our findings reveal distinct metabolic responses to Aβ toxicity, with significant alterations in pyruvate and glutamine metabolism. Both GAL and LYC contributed to the restoration of glutamine and lysine homeostasis, but LYC had a more pronounced effect, better sustaining cellular energy balance and mitochondrial function. Unlike LYC, GAL treatment was associated with pyruvate accumulation, highlighting a distinct metabolic response between the two compounds. These variations may reflect distinct mechanisms of action, potentially influencing their therapeutic roles in counteracting Aβ-induced toxicity. This study highlights the value of metabolic profiling for assessing neuroprotective agents and reinforces the potential of natural alkaloids in this context.

Background/Objectives: Natural products containing hydroxyanthracene derivatives (HADs) such as Cascara (Rhamnus purshiana), Frangula (Rhamnus frangula), Rhubarb (Rheum palmatum), and Senna (Cassia angustifolia) have long been used for their laxative properties, but also raise safety concerns due to reported genotoxic and carcinogenic potential. Most studies have focused on quantifying HADs, whereas the broader secondary metabolite landscape of these herbal drugs remains underexplored. We aimed to generate an untargeted metabolomic fingerprint of these four species and to explore their chemical diversity using AI-based structural classification. Methods: Four commercial botanical raw materials were extracted with 60% methanol and analysed by UPLC–HRMS/MS in positive and negative ion modes. Features were processed in Compound Discoverer and annotated by accurate mass and MS/MS matching against spectral databases, then assigned to structural classes using a graph neural network classifier. Multivariate analyses (PCA, HCA) were used to compare metabolic patterns across species. Results: In total, 93, 83, 83 and 51 metabolites were annotated in cascara, frangula, rhubarb, and senna, respectively, spanning flavonoids, anthraquinones, phenylpropanoids and other classes. Only four flavonoids were shared by all species, indicating marked biochemical divergence. Several putatively species-enriched features were observed, including pavine in cascara and frangula, vicenin-2 in senna, and piceatannol in rhubarb. Senna displayed the most distinct metabolic profile, whereas cascara and frangula clustered closely. Conclusions: This work provides a chemistry-centred metabolomic fingerprint of four HAD-containing herbal drugs using graph-based neural networks for natural product classification, supporting future studies on the pharmacological potential, bioavailability and safety of their metabolites.

Alzheimer’s Disease (AD) affects the quality of life of millions of people worldwide and represents one of the biggest challenges for the whole society. The SH-SY5Y neuroblastoma cell line is often used as an in vitro model of neuronal function and is widely applied to study the molecular events leading to AD. In the last few years, basic research on SH-SY5Y cells has provided interesting insights for the discovery of new drugs and biomarkers for improved AD treatment and diagnosis. At the same time, untargeted NMR metabolomics is widely applied on biological fluids for (i) metabolic profile analysis, (ii) screening for differential metabolites, (iii) analysis of metabolic pathways, and (iv) the discovery of new biomarkers. Principal Component Analysis (PCA) and Linear Discriminant Analysis (LDA) have proved to be powerful methods for processing NMR data, being useful in signal quantization, even if more sophisticated --- typically non--linear --- techniques are needed to obtain compact yet information--rich embeddings for complex spectra. In this paper, a compression technique based on convolutional autoencoders is proposed, which can perform a high dimensionality reduction of the spectral signal (up to more than 300 times), maintaining informative features (guaranteed by a reconstruction error always smaller than 5%). Moreover, before compression, an ad hoc preprocessing method was devised to remedy the scarcity of available data. The compressed spectral data were then used to train some SVM classifiers to distinguish diseased from healthy cells, achieving an accuracy close to 78%, a significantly better performance with respect to using PCA--compressed data.

Continuous monitoring of climatic variables is essential for precision viticulture and data-driven decision support systems. However, agricultural sensor networks are frequently affected by missing data due to hardware failures, communication issues, or maintenance interruptions. In this work, we propose a spatio-temporal graph-based autoencoder for reconstructing missing temperature and relative humidity time series collected from a five-node vineyard sensor network over a two-year period. The model combines a GRU-based temporal encoder, augmented with a time-decay imputation mechanism applied to the input data, with a GraphSAGE spatial module, enabling the joint exploitation of temporal dynamics and inter-node spatial correlations. Experimental results on real-world data show that the proposed approach achieves accurate reconstruction under controlled missing-data scenarios generated through structured artificial masking. For moderate corruption levels ($p=0.3$), the model attains reconstruction losses of $0.003$ for temperature and $0.005$ for humidity using short temporal windows (L = 36~3 h), corresponding to MAE values below 0.03 °C and 0.1%, respectively. Even at higher corruption levels ($p=0.7$), performance remains stable, with losses below $0.008$ and $0.011$, and MAE values within 0.05 °C and 0.17%. The results highlight a trade-off between temporal context and reconstruction accuracy: shorter windows yield lower absolute errors under moderate corruption whereas, under extreme data loss ($p=0.9$), the longer windows (L = 144~12 h) reduce the composite temperature reconstruction loss from 0.027 to 0.021. Additionally, temperature is consistently reconstructed more accurately than humidity, reflecting its smoother dynamics and stronger spatial coherence.

Continuous monitoring of microclimatic variables is essential for precision viticulture and data-driven decision support systems. However, agricultural sensor networks are frequently affected by missing data due to hardware failures, communication issues, or maintenance interruptions. In this work, we propose a spatio-temporal graph-based autoencoder for reconstructing missing temperature and relative humidity time series collected from a five-node vineyard sensor network over a two-year period. The model combines a GRU-D-based temporal encoder with a GraphSAGE spatial module, enabling the joint exploitation of temporal dynamics and inter-node spatial correlations. Experimental results on real-world data show that the proposed approach achieves accurate reconstruction under realistic missing-data conditions. For moderate corruption levels (p=0.3), the model attains reconstruction losses of 0.003 for temperature and 0.005 for humidity using short temporal windows (L=36∼3h), corresponding to MAE values below 0.03∘C and 0.1%, respectively. Even at higher corruption levels (p=0.7), performance remains stable, with losses below 0.008 and 0.011, and MAE values within 0.05∘C and 0.17%. The results highlight a trade-off between temporal context and reconstruction stability: shorter windows yield better performance under moderate corruption, while longer windows (L=144∼12h) improve robustness under extreme data loss (p=0.9), reducing temperature reconstruction loss from 0.027 to 0.021 and MAE from 0.133∘ to 0.226∘. Additionally, temperature is consistently reconstructed more accurately than humidity, reflecting its smoother dynamics and stronger spatial coherence.