Background: Pepsin, a component of gastric refluxate, has been investigated as a potential salivary biomarker for gastro-esophageal reflux disease (GERD), although its diagnostic accuracy remains uncertain. This proof-of-concept study aimed to characterize the effects of pepsin on the human salivary peptidome using matrix-assisted laser desorption/ionization–time of flight (MALDI-ToF) mass spectrometry. Methods: Whole saliva samples were collected from ten healthy adult volunteers under fasting conditions and divided into untreated controls and aliquots digested with pepsin at acidic pH. MALDI-ToF MS was used to profile digestion-induced changes in peptide mass patterns. Spectral data were analyzed using multivariate statistical approaches, including principal component analysis (PCA), linear discriminant analysis (LDA), and hierarchical clustering. Results: Pepsin digestion increased peptide signal intensity and spectral complexity compared with controls. PCA demonstrated clear separation between native and digested samples along the first principal component. Ten peptide m/z features showed the strongest association with pepsin exposure based on PCA loadings. LDA and hierarchical clustering further supported this distinction, with the top 15 discriminative m/z features showing consistent enrichment in digested samples despite inter-individual variability. Conclusions: Pepsin exposure induces reproducible remodeling of the salivary peptidome detectable by MALDI‑ToF MS. Although this peptide‑level approach cannot resolve the full diversity of salivary proteoforms, the resulting signatures support the feasibility of identifying markers of reflux‑associated enzymatic activity and provide a basis for future validation in clinical GERD cohorts.

Wearable electrochemical biosensors often producevoltammetric signals that are corrupted by noise and long-term drift.Effective on-device denoising is critical to improve signal quality anddetect anomalies due to sensor drift or interference. This paperexplores lightweight TinyML models for denoising and drift detectionin wearable sensor voltammograms under the strict memoryconstraints of microcontrollers. We apply compact 1D convolutionaland dense autoencoder networks, as well as a PCA-basedreconstruction, to remove noise and identify drifting signals. Using apublic NIST dataset of cyclic voltammograms with added syntheticnoise and artifacts, we evaluate each model’s denoising performance(signal reconstruction MSE) and drift/anomaly detection capability (ROC-AUC) versus its memory footprint (quantized int8 model size). Results show that a small Conv1D autoencoder (8KB weights) canreduce noise by 75% and achieve 0.89 AUC for drift detection,approaching the performance of a larger dense autoencoder (35KB)and outperforming PCA. We observe a trade-off between model sizeand generalization: the larger autoencoder nearly perfectly flaggedanomalies (AUC 1.0) but smaller models remain competitive whileusing 4–6× less memory. These findings demonstrate that drift-resilient signal enhancement can be achieved on-device with minimalresource usage, enabling more robust wearable electrochemicalsensing.