Article
Biochemistry & Molecular Biology
Mir Henglin, Brian L. Claggett, Joseph Antonelli, Mona Alotaibi, Gino Alberto Magalang, Jeramie D. Watrous, Kim A. Lagerborg, Gavin Ovsak, Gabriel Musso, Olga V. Demler, Ramachandran S. Vasan, Martin G. Larson, Mohit Jain, Susan Cheng
Summary: This study compares the performance of traditional and newer statistical learning methods in different types of metabolomics data and finds that, in the analysis of human metabolomics data, multivariate methods perform better than univariate methods as the number of study subjects increases.
Review
Biochemistry & Molecular Biology
Janet Blaurock, Sven Baumann, Sonja Grunewald, Juergen Schiller, Kathrin M. Engel
Summary: This review summarizes the attempts made so far to overcome the decreasing fertilizing ability of human spermatozoa using metabolomics-based investigations. Various analytical methods, including liquid chromatography-mass spectrometry (LC-MS), gas chromatography (GC), infrared (IR), Raman, and nuclear magnetic resonance (NMR) spectroscopy, have been discussed. Results show that LC-MS detects the highest number of metabolites and can be considered as the method of choice. However, the reproducibility of some studies is poor, and further improvements of study designs are needed, along with a stronger focus on the biochemical consequences of altered metabolite concentrations.
INTERNATIONAL JOURNAL OF MOLECULAR SCIENCES
(2022)
Article
Genetics & Heredity
Osval A. Montesinos-Lopez, Abelardo Montesinos-Lopez, Bernabe Cano-Paez, Carlos Moises Hernandez-Suarez, Pedro C. Santana-Mancilla, Jose Crossa
Summary: Genomic selection has revolutionized the way plant breeders select genotypes, using statistical machine learning models to predict phenotypic values of new lines. Multi-trait genomic prediction models leverage correlated traits to improve accuracy. This paper compares the performance of three multi-trait methods and finds that their performance varies under different predictors.
Article
Computer Science, Artificial Intelligence
Niklas Fries, Patrik Ryden
Summary: This study investigates the use of local explanation methods for root cause analysis in high-dimensional industrial applications. Results show that generalized linear methods provide the best explanations for simulated data, while TreeExplainer performs similarly for real-world industrial data. Different explanation methods vary in performance across different relationships.
EXPERT SYSTEMS WITH APPLICATIONS
(2022)
Article
Environmental Sciences
He Li, Peng Wang, Chong Huang
Summary: This paper compares the performance of three common DL algorithms (EfficientDet, SSD, and YOLOv4) for sorghum head detection based on lightweight UAV remote sensing data. The study found that YOLOv4 had the most accurate detection results, while EfficientDet performed the worst. Additionally, the overlap ratio, confidence, and IoU parameters also influenced the detection results.
Article
Marine & Freshwater Biology
Lorrana Thais Maximo Durville Braga, Alejandro Giraldo, Alexandre Lima Godinho
Summary: The study compared three methods of counting fish manually in the frames of DIDSON software in two Brazilian dams, with qualitative counting being the most congruent with numerical counting.
Article
Materials Science, Multidisciplinary
Annelies De Wael, Annick De Backer, Chu-Ping Yu, Duygu Gizem Sentuerk, Ivan Lobato, Christel Faes, Sandra Van Aert
Summary: This paper introduces a statistical method to count the number of atoms in ADF STEM images and discusses how to represent the counting results together with their statistical error. Three approaches are presented and their suitability for further use is evaluated based on simulations and experimental images.
MICROSCOPY AND MICROANALYSIS
(2023)
Article
Biology
Christelle Damon-Soubeyrand, Antonino Bongiovanni, Areski Chorfa, Chantal Goubely, Nelly Pirot, Luc Pardanaud, Laurence Piboin-Fragner, Caroline Vachias, Stephanie Bravard, Rachel Guiton, Jean-Leon Thomas, Fabrice Saez, Ayhan Kocer, Meryem Tardivel, Joel R. Drevet, Joelle Henry-Berger
Summary: The epididymis plays a crucial role in male fertility and has both secretory and immune functions. However, the organization of blood and lymphatic networks in the epididymis is still largely unknown.
Article
Astronomy & Astrophysics
Noah Weaverdyck, Dragan Huterer
Summary: The text discusses the meticulous treatment of systematic contamination in large-scale structure surveys, comparing existing methods for removing systematics from galaxy clustering measurements and suggesting improved estimators. It also explores methods to generate clean maps and reduce errors in power spectra, proposing two new mitigation methods. The performance of these methods is tested on simulated measurements, resulting in improved maps and power spectra with minimal user tuning, providing recommendations for systematics mitigation in future surveys.
MONTHLY NOTICES OF THE ROYAL ASTRONOMICAL SOCIETY
(2021)
Article
Environmental Sciences
Mara Meggiorin, Giulia Passadore, Silvia Bertoldo, Andrea Sottani, Andrea Rinaldo
Summary: This study compares three imputation methods for filling missing data in groundwater elevation timeseries observations. The methods tested are spline interpolation, autoregressive linear model, and patched kriging. The results show that spline interpolation performs poorly, while patched kriging proves to be the best option, despite occasional false trends. The choice of imputation method has minimal effect on the overall statistics.
Article
Multidisciplinary Sciences
Jeonggil Lee, Han-Suk Kim, Ho Young Jo, Man Jae Kwon
Summary: Different methods have been explored for quantifying soil bacteria, with challenges in identifying the optimal approach. Storage temperature and pretreatment methods play a crucial role in preserving bacterial viability and reducing interference from soil particles. Among the counting methods evaluated in this study, the most probable number (MPN) was rapid and reliable, but may underestimate anaerobic bacteria. DNA quantitation method tended to overestimate bacterial numbers.
Article
Chemistry, Analytical
Athanasios Lentzas, Eleana Dalagdi, Dimitris Vrakas
Summary: As the population ages and access to ambient sensors becomes easier, activity recognition in smart home installations has gained increased scientific interest. While most previous studies focused on recognizing activities of single residents, this study investigates activity recognition for multiple residents concurrently, treating it as a multilabel classification problem. Experimental comparison of different algorithms showed that using multilabel classification can accurately recognize activities performed by multiple people.
Article
Environmental Sciences
Barbara Barzycka, Mariusz Grabiec, Jacek Jania, Malgorzata Blaszczyk, Finnur Palsson, Michal Laska, Dariusz Ignatiuk, Gudfinna Adalgeirsdottir
Summary: This study uses a unique dataset including satellite SAR images, Ground Penetrating Radar data, and shallow glacier cores to assess the performance of different methods in distinguishing glacier zones. The findings suggest that unsupervised classification methods, such as Gaussian Mixture Model-Expectation Maximization (GMM-EM) clustering of dual-pol backscattering coefficient (sigma0) and quad-pol Pauli decomposition, show promise in distinguishing glacier zones.
Article
Chemistry, Multidisciplinary
Xin Zhang, Yanqiao Zhao, Bin Li, Meichen Guo, Jinwu Lv, Yuantao Wei
Summary: The effects of three extraction methods (ultrasonic-assisted water extraction, ultrasonic-assisted ethanol extraction, and ultrasonic-assisted aqueous two-phase extraction) on the yields, physicochemical properties, stabilities, and antioxidant activities of anthocyanins from perilla leaves were investigated. The results showed that ultrasonic-assisted aqueous two-phase extraction yielded the highest content of anthocyanins, with the best color quality, stability, and antioxidant activities.
SUSTAINABLE CHEMISTRY AND PHARMACY
(2022)
Article
Multidisciplinary Sciences
Vajira Thambawita, Steven A. Hicks, Andrea M. Storas, Thu Nguyen, Jorunn M. Andersen, Oliwia Witczak, Trine B. Haugen, Hugo L. Hammer, Pal Halvorsen, Michael A. Riegler
Summary: A manual assessment of sperm motility using microscopy observation is challenging due to the fast movement of spermatozoa. Therefore, computer-aided sperm analysis (CASA) has become more common in clinics. However, more data is needed to train supervised machine learning models for accurate evaluation. In this study, a dataset called VISEM-Tracking is provided, which includes annotated video recordings of wet semen preparations and experts' analysis of sperm characteristics. The dataset can be used to train deep learning models for sperm analysis.