Food Analytical Methods最新文献

{"title":"Discrimination of Chinese Baijiu Base Liquor Grades Using Near-Infrared Spectroscopy Based on SHAP and Multi-scale Pyramid Feature Fusion","authors":"Xianguo Tuo,&nbsp;Longxin Li,&nbsp;Guiyu Zhang,&nbsp;Kerong Chen,&nbsp;Xianglin Zeng","doi":"10.1007/s12161-026-03236-z","DOIUrl":"10.1007/s12161-026-03236-z","url":null,"abstract":"<div><p>To address the challenges of high-dimensional redundancy in near-infrared spectroscopy (NIRS) for grading Chinese Baijiu base liquor and to overcome the limitations of traditional feature selection methods—which often lose local spectral shape information due to an excessive focus on discrete wavelengths—this study proposes a dual-feature optimization framework. The methodology initiates by circumventing conventional distance metrics through a graph-based splitting strategy that leverages spectral clustering to characterize the manifold structure of samples within the feature space. Following this, a \"point-plane\" fusion mechanism is introduced: SHAP (Shapley Additive Explanations) values derived from an eXtreme Gradient Boosting (XGBoost) model are utilized to quantify the global contribution of each wavelength, thereby facilitating the selection of key discrete wavelengths (points). Simultaneously, multi-scale pyramid features are incorporated to extract local texture and shape information (representing the 'planes') from the spectra via hierarchical statistical pooling, resulting in a more refined representation of NIRS data. Ultimately, the XGBoost model is implemented for classifying base liquor grades. On the graph-based hold-out partition, the final SHAP–Pyramid–XGBoost model achieved an Accuracy of 0.9688 and a Macro-F1 score of 0.9696. Repeated nested stratified five-fold cross-validation with five repeats yielded a more conservative Accuracy of 0.9083 ± 0.0237 and a Macro-F1 score of 0.9063 ± 0.0241. Evaluation on three public spectroscopic benchmark datasets further indicated that the contribution of pyramid fusion was dependent on the dataset and wavelength-selection method. These results support promising internal discrimination performance on the present Baijiu dataset, while independent multi-site validation remains necessary.</p></div>","PeriodicalId":561,"journal":{"name":"Food Analytical Methods","volume":"19 10","pages":""},"PeriodicalIF":3.0,"publicationDate":"2026-09-02","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"148926293","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":3,"RegionCategory":"农林科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Optimization of Pectin Extraction from Yellow Hawthorn (Crataegus azarolus) and Initial Evaluation of Its Potential for Eco-Friendly Biopolymer Films","authors":"Amina Menasra,&nbsp;Ouarda Amrani,&nbsp;Imane Hireche","doi":"10.1007/s12161-026-03235-0","DOIUrl":"10.1007/s12161-026-03235-0","url":null,"abstract":"<div><p><i>Crataegus azarolus</i> (Hawthorn) serves as a novel, untapped botanical source for pectin extraction, offering a valuable approach to valorize underutilized wild fruits. This study optimized pectin extraction yield and quality using 0.1 M citric acid by varying pH and temperature, and comprehensively evaluated physicochemical, functional, and structural properties to assess its preliminary potential for preparing eco-friendly packaging films. Extractions were performed across pH levels of 1.5, 2.5, and 3.5, and temperatures of 70, 80, and 90 °C. Results showed that optimal extraction conditions yielded a maximum (8.43%) at pH 1.5 and 80 °C, whereas milder conditions (pH 3.5 at 70 °C) successfully preserved the highest degree of esterification (67.01%) and methoxyl content (9.22%). Extracted pectins exhibited high purity with low ash content (0.73–1.49%), excellent functional properties, and characteristic FTIR profiles confirming typical structural features, such as esterified carboxyl bands around 1740 cm⁻¹ and ionized carboxyl bands near 1600 cm⁻¹. Response Surface Methodology revealed significant quadratic models (p &lt; 0.05) with high R² values (&gt; 0.91), confirming strong agreement between experimental and predicted values. Temperature significantly affected yield and color parameters, while pH strongly influenced structural and functional behavior. Film-forming solutions exhibited shear-thinning behavior with viscosities ranging from 280.7 to 1000.05 mPa·s, and cast films showed uniform thicknesses of 0.072–0.089 mm. Although these rheological and physical traits are promising, further characterization, including mechanical testing, is required before establishing viability for commercial packaging applications. These findings highlight <i>Crataegus azarolus</i> as a renewable biopolymer source for future sustainable development.</p><h3>Graphical Abstract</h3>\u0000<div><figure><div><div><picture><source><img></source></picture></div></div></figure></div></div>","PeriodicalId":561,"journal":{"name":"Food Analytical Methods","volume":"19 10","pages":""},"PeriodicalIF":3.0,"publicationDate":"2026-09-02","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"148926292","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":3,"RegionCategory":"农林科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Fluorescence Spectroscopy Coupled with Artificial Neural Networks to Detect Refined Palm Oil Adulteration in Extra Virgin Olive Oil: A Case Study on Commercialized Oils in Cameroon","authors":"William Mbogning Feudjio,&nbsp;Ikram Khemiri,&nbsp;Aymar Rodrigue Fogang Mba,&nbsp;Sagesse Bel Christ Kogniwali-Gredibert,&nbsp;Gabrielle Muke Ngome,&nbsp;Lionel José Wankam,&nbsp;Gilbert Yvon Mbesse Kongbonga,&nbsp;Cyril Kenfack Assongo,&nbsp;Hassen Ghalila,&nbsp;Mama Nsangou","doi":"10.1007/s12161-026-03250-1","DOIUrl":"10.1007/s12161-026-03250-1","url":null,"abstract":"<div><p>The adulteration of extra virgin olive oil remains a major problem for the food industry. This practice has both health and economic consequences. In countries which import extra virgin olive oil, the problem is even more serious. The objective of this study was to use a laboratory assembled fluorometer to detect the presence of refined palm oil in extra virgin olive oil commercialized in Cameroon. The fluorescence spectra obtained with an excitation wavelength of 380 nm reveal two principal emission peaks: one at 497.18 nm, attributed to oxidation products in refined palm oil, and another at 673.75 nm, attributed to chlorophylls in extra virgin olive oil. The fluorescence intensities of the adulterated samples increased with the concentration of refined palm oil in extra virgin olive oil. Principal component analysis, employing two principal components that accounted for 95% of the total variance, enabled clear visualization of the clusters of adulterated samples. The risk of misclassification between classes, as indicated by principal component analysis, was mitigated by a self-organizing maps neural network. Finally, a multilayer perceptron neural network demonstrated high efficacy in predicting with a determination coefficient of 0.998, the concentration of refined palm oil in extra virgin olive oil. To the best of our knowledge, this study represents the first application of fluorescence spectroscopy coupled with artificial neural networks to detect the adulteration of extra virgin olive oil with refined palm oil. The results obtained prove the significance of fluorescence and neural network tools for the reliable characterization of vegetable oils, opening promising perspectives for their application in the food industry.</p></div>","PeriodicalId":561,"journal":{"name":"Food Analytical Methods","volume":"19 10","pages":""},"PeriodicalIF":3.0,"publicationDate":"2026-09-02","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"148926294","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":3,"RegionCategory":"农林科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Study on the Mechanism and Application of Enzyme-modified Starch for Efficient Recovery of Salmonella in Beef","authors":"Yuxin Yang,&nbsp;Mei Shu,&nbsp;Ningbo Liao,&nbsp;Chan Zhong,&nbsp;Guoping Wu","doi":"10.1007/s12161-026-03242-1","DOIUrl":"10.1007/s12161-026-03242-1","url":null,"abstract":"<div><p>The colloidal properties of <i>Salmonella</i> promote its adhesion to the oil–water interface in fatty food emulsions, hindering separation from the aqueous phase and reducing the sensitivity of rapid detection assays. There have been studies that utilized α-amylase to hydrolyze starch to enhance the detection of <i>Salmonella</i> in ground beef. However, the interfacial behavior of enzyme-modified starch and its relationship with bacterial distribution in oil–water systems is unclear. Based on this background, an enzyme-modified soluble starch (E-SS) was prepared that can effectively recover <i>Salmonella</i> from meat samples. Furthermore, this study examined the molecular structure of E-SS and its regulatory effect on <i>Salmonella</i> distribution in oil–water systems. Compared with native soluble starch (SS), E-SS exhibited a smaller molecular weight distribution (Mw = 3.184 × 10³ ± 5.776%) and stronger interfacial activity (<i>K</i>_<i>diff</i> = 1.14 ± 0.02). Confocal laser scanning microscopy (CLSM) revealed fluorescence co-localization between E-SS and the oil phase, indicating that E-SS was uniformly adsorbed in the oil–water interface. The better interfacial properties of E-SS allowed it to competitively displace <i>Salmonella</i> adsorbed on the oil–water interface and facilitate the enrichment of the bacteria in the aqueous phase. Finally, the early detection methods were optimized and iterated: E-SS was combined with Bentonite-coated activated carbon (BCAC) and real-time fluorescence loop-mediated isothermal amplification (Rti-LAMP) to detect <i>Salmonella</i> in beef samples. It was found that the addition of E-SS improved the detection sensitivity of BCAC-Rti-LAMP for <i>Salmonella</i> by 10 to 100-fold within 3 h.</p></div>","PeriodicalId":561,"journal":{"name":"Food Analytical Methods","volume":"19 10","pages":""},"PeriodicalIF":3.0,"publicationDate":"2026-09-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"148926565","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":3,"RegionCategory":"农林科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Electric Field and Pressure Assisted Extraction of Mushkbudji Rice Bran Oil: Optimization, Structural and Antioxidant Characterization and Comparative Evaluation of Soxhlet Extraction","authors":"Nusrat Bhat,&nbsp;Nusrat Maqbool,&nbsp;Thameed Aijaz","doi":"10.1007/s12161-026-03234-1","DOIUrl":"10.1007/s12161-026-03234-1","url":null,"abstract":"<div><p>The present study investigates a novel electric field and pressure assisted extraction (EF&amp;PAE) technique for the recovery of rice bran oil (RBO) from the geographically indicated Mushkbudji rice (<i>Oryza sativa</i> L) compared to conventional Soxhlet extraction. Process variables like electric field strength (10–30 V/cm), positive pressure (0–10 kg/cm²), and temperature (10–20 ℃) were evaluated for their effect on the efficiency of extraction. The maximum oil recovery (16.01 g/100 g) was achieved at 30 V/cm, 10 kg/cm², and 20 ℃, representing a 33.4% improvement over Soxhlet extraction (12.00 g/100 g). Physicochemical properties indicated that EF&amp;PAE RBO was of better quality, showing significantly lower FFA (free fatty acid) (5.41%), peroxide value (4.486 meq O₂/kg), and thiobarbituric acid value (2.88 mg MDA/kg) than Soxhlet extracted RBO. Antioxidant profiling revealed significantly higher total flavonoid content (13,124.5 mg QE/100 g), total phenolic content (29.384 mg GAE/g), and ABTS radical scavenging (63.00%) in EF&amp;PAE-extracted RBO, with similar FRAP values. But Soxhlet RBO exhibited superior DPPH scavenging activity (lower IC₅₀), revealing assay-specific antioxidant activity. FTIR confirmed triglyceride integrity in both oils, with greater extraction of lipid component in EF&amp;PAE, and absence of degradation. EF&amp;PAE proves to be a fast, low-temperature and eco-friendly method for extraction, yielding antioxidant-rich, oxidatively stable oil. This is the first study to report EF&amp;PAE application in Mushkbudji rice bran, showing its capability for nutraceutical oil extraction.</p></div>","PeriodicalId":561,"journal":{"name":"Food Analytical Methods","volume":"19 10","pages":""},"PeriodicalIF":3.0,"publicationDate":"2026-09-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"148926566","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":3,"RegionCategory":"农林科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Aptasensors for Rapid Melamine Detection in Dairy Foods: Recent Developments and Trends","authors":"Veena Nagarajappa,&nbsp;K S Chandana Suresh,&nbsp;Sivakumar Shanmugam","doi":"10.1007/s12161-026-03244-z","DOIUrl":"10.1007/s12161-026-03244-z","url":null,"abstract":"<div><p>Melamine (MA) detection in milk and milk products has attracted much attention since the discovery that MA-adulterated food causes serious damage to the kidney, leading to urinary calculus, acute kidney failure, and bladder cancer. Most methods measure MA through various chromatographic techniques, which require skill, time consuming, tedious, expensive instrumentation and preprocessing of samples. Aptamer-based biosensors are a promising choice for MA analysis due to their advantages over instrumental analysis and immunoassays. Thymine (T) in single-stranded DNA can specifically bind to MA through a T-MA-T triple hydrogen bonding (NH⋯O and NH⋯N) motif, which serves as the fundamental recognition principle for many DNA-based biosensors. In this review, we aim to report the latest research progress in the field of aptasensors based on different sensing technologies (including colorimetric, fluorescence, surface enhanced Raman scattering (SERS), resonance Rayleigh scattering (RRS), and electrochemical) for rapid detection of MA in milk products. While these sensing techniques enable accurate and rapid detection, portable, high sensitivity, high efficiency, and simple operation, they are constrained by matrix interference, limited stability, poor reproducibility, and insufficient standardization. Overcoming these hurdles to achieve large-scale implementation requires future research to enhance device robustness, selectivity, and data-processing capabilities. This review will inspire future research on aptamer-based biosensing for real time monitoring of MA adulteration in order to safeguard human health and food safety.</p></div>","PeriodicalId":561,"journal":{"name":"Food Analytical Methods","volume":"19 10","pages":""},"PeriodicalIF":3.0,"publicationDate":"2026-09-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"148926495","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":3,"RegionCategory":"农林科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Response Surface Optimization of Lavatera cretica Leaves Extract: HPLC–UV Phenolic Profiling, Antioxidant and Anti-hemolytic Activities","authors":"Asma Ourabah,&nbsp;Fares Boudjouan,&nbsp;Walid Zeghbib,&nbsp;Farah Yous,&nbsp;Ali Benadjaoud,&nbsp;Nadjet Debbache-Benaida","doi":"10.1007/s12161-026-03239-w","DOIUrl":"10.1007/s12161-026-03239-w","url":null,"abstract":"<div><p><i>Lavatera cretica</i> leaves were traditionally used as a medicinal and food plant, particularly in rural communities. This study aimed to optimize extraction conditions for phenolic compounds from <i>Lavatera cretica</i> leaves and to evaluate their antioxidant and anti-hemolytic activities. The Box-Behnken model was used to optimize maceration time, solid-to-liquid ratio, and ethanol concentration. High-performance liquid chromatography with ultraviolet detection (HPLC–UV) was used to analyze the bioactive compounds in the extract. In vitro antioxidant and anti-hemolytic assays were also conducted. The optimal extraction conditions were 46.18% (v/v) ethanol, 3.42 g/50 mL, and 08 h. The extract was rich in bioactive compounds, with 85.58 ± 2.82 mg GAE/g DE for total phenols, 32.99 ± 0.30 mg QE/g DE for flavonoids, and 51.13 ± 4.67 mg CE/g DE for tannins. HPLC–UV analysis revealed a diverse phenolic profile, with rutin (0.478 mg eq/g DW) and p-coumaric acid (0.476 mg eq/g DW) among the major quantified constituents. The extract exhibited effective antioxidant activity, with a ferric reducing power of 166.30 ± 1.86 mg TE/g DE, and inhibition of 70.23 ± 2.96% and 27.64 ± 0.12% against 2,2-Diphenyl-1-picrylhydrazyl (DPPH) and hydrogen peroxide (H2O2), respectively. In addition, an anti-hemolytic effect was observed at 500 µg/mL, with inhibition of 98.07 ± 0.03%. The present study provides insight into the biological potential of <i>Lavatera cretica</i> leaves for pharmacological applications in human health.</p></div>","PeriodicalId":561,"journal":{"name":"Food Analytical Methods","volume":"19 10","pages":""},"PeriodicalIF":3.0,"publicationDate":"2026-08-31","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"148928030","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":3,"RegionCategory":"农林科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"From Kitchen to Classroom: Do-It-Yourself (DIY) Natural Sensors for Everyday Health and Food Quality Monitoring","authors":"Suvendu Paul,&nbsp;Monaj Karar,&nbsp;Nilanjan Dey","doi":"10.1007/s12161-026-03227-0","DOIUrl":"10.1007/s12161-026-03227-0","url":null,"abstract":"<div><p>This study explores the design and application of red cabbage-derived DIY sensors as low-cost, eco-friendly analytical tools for monitoring quality parameters in food safety, personal health, and environmental diagnostics. Red cabbage (<i>Brassica oleracea</i> var. <i>capitata</i> f. <i>rubra</i>) is a rich source of anthocyanins, naturally occurring pH-sensitive pigments that undergo distinct, reversible color transitions across a pH range of 2–10. Leveraging these chromogenic shifts, we developed biodegradable sensors for monitoring sweat pH, exhaled breath gas, and food quality, fabricated from household materials such as filter paper, gelatin, and vinegar. These sensors demonstrated high visual contrast and functional reliability when exposed to bio-mimicking stimuli. To assess the pedagogical utility of these sensors, interactive sessions were conducted with high school and undergraduate students (<i>n</i> = 30), integrating sensor fabrication, calibration, and smartphone-assisted RGB analysis. Pre- and post-session assessments revealed a 38–86% increase in conceptual understanding related to acid–base chemistry, anthocyanin structure–function relationships, and sustainability in materials chemistry. Fabrication accuracy and color calibration performance exceeded 70% across groups, with strong student engagement (Likert score ≥ 4.6) and a reported increase in confidence for conducting independent home experiments. These results affirm the scientific robustness and educational value of red cabbage-based colorimetric sensors, offering a scalable approach to democratize analytical chemistry through hands-on, real-world experimentation in resource-limited and classroom environments.</p><h3>Graphical Abstract</h3>\u0000<div><figure><div><div><picture><source><img></source></picture></div></div></figure></div></div>","PeriodicalId":561,"journal":{"name":"Food Analytical Methods","volume":"19 10","pages":""},"PeriodicalIF":3.0,"publicationDate":"2026-08-30","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"148928004","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":3,"RegionCategory":"农林科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Adulteration Detection in Lemon Essential Oil: Comparison of Different Spectroscopy Techniques and Machine Learning Applications","authors":"Qinghua Liu,&nbsp;Ziyang Wu,&nbsp;Huihui Yang,&nbsp;Qian Su,&nbsp;Minmin Li,&nbsp;Bei Fan,&nbsp;Fengzhong Wang,&nbsp;Long Li","doi":"10.1007/s12161-026-03247-w","DOIUrl":"10.1007/s12161-026-03247-w","url":null,"abstract":"<div><p>Essential oil adulteration severely compromises product quality and market credibility. This study proposes a rapid, nondestructive approach combining optical spectroscopy with chemometrics to quantify lemon essential oil adulterated with sweet orange essential oil. Three spectroscopic techniques—Fourier transform infrared (FT-IR), three-dimensional fluorescence (3D-FD), and portable near-infrared (po-NIR) spectroscopy—were systematically evaluated. Spectral data were optimized using various preprocessing algorithms, followed by characteristic wavelength selection via competitive adaptive reweighted sampling (CARS) and quantification modelling using partial least squares regression (PLSR). Results demonstrated that the FT-IR model coupled with standard normal variate and CARS-PLSR achieved optimal predictive performance, yielding a validation coefficient of determination (Rp²) of 0.9998 and a root mean square error of prediction of 0.4180. The 3D-FD technique also exhibited exceptional accuracy (Rp² = 0.9994). Furthermore, despite susceptibility to environmental noise, the po-NIR model optimized by normalization-CARS-PLSR delivered satisfactory on-site detection capability (Rp² = 0.9663). These findings highlight that integrating advanced spectroscopic techniques with robust machine learning algorithms provides a highly reliable strategy for essential oil quality control, establishing a solid foundation for portable onsite instrumentation.</p></div>","PeriodicalId":561,"journal":{"name":"Food Analytical Methods","volume":"19 10","pages":""},"PeriodicalIF":3.0,"publicationDate":"2026-08-29","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"148927877","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":3,"RegionCategory":"农林科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"A Time Series Modeling Method for Soft Sensing and Prediction of Daqu Acidity Based on Environmental Responses","authors":"Haili Yang,&nbsp;Hao Xia,&nbsp;Jing Yang,&nbsp;Xinjun Hu,&nbsp;Manjiao Chen,&nbsp;Dan Huang,&nbsp;Liangliang Xie,&nbsp;Fuhao Han,&nbsp;Yuqi Zhou,&nbsp;Suyi Zhang,&nbsp;Ping Yang,&nbsp;Wanghui Su","doi":"10.1007/s12161-026-03255-w","DOIUrl":"10.1007/s12161-026-03255-w","url":null,"abstract":"<div><p>Daqu is a key fermentation starter for Baijiu production, whose acidity is a core indicator for evaluating its quality. This study aimed to propose a method for Daqu acidity soft sensing and time series prediction based on environmental response so as to address the issues of complexity, destructiveness, and lag associated with traditional detection methods. The environmental parameters employed during Daqu fermentation were recorded and preprocessed. A stacking ensemble soft sensing model based on gradient boosting decision tree, random forest, extreme gradient boosting (XGBoost), and ridge regression was constructed to achieve rapid, nondestructive, and real-time estimation of Daqu acidity. The hourly acidity estimates output by the soft sensing model were then used as input features, besides environmental parameters, to construct a hybrid XGBoost–LSTM time series prediction model for predicting future acidity values. The experimental results showed that the stacking soft sensing model achieved the best performance, with a coefficient of determination (<i>R</i>²) of 0.9409. Compared with a single XGBoost model, the XGBoost–LSTM time series model improved <i>R</i>² by more than 5% on average across various monitoring points, with a maximum improvement of 9.96%. The proposed method was validated using data from different fermentation rooms and independent batches, demonstrating strong generalization ability and robustness. This study provides an efficient, accurate, and nondestructive technical support for intelligent monitoring of the Daqu fermentation process.</p></div>","PeriodicalId":561,"journal":{"name":"Food Analytical Methods","volume":"19 10","pages":""},"PeriodicalIF":3.0,"publicationDate":"2026-08-29","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"148928113","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":3,"RegionCategory":"农林科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}