Results 241 to 250 of about 201,162 (313)

Comparative classification of spectrally overlapping Allium seed genotypes using Vis-NIR spectroscopy and hyperspectral imaging with chemometric, machine, and deep learning models. [PDF]

Sci Rep
Fathi G, Mireei SA, Jafari M, Sadeghi M, Karimmojeni H, Nazeri M, Tylewicz U.   +6 more
europepmc   +1 more source

Robustness of Llama3 F1 Scores Under Mixed Data Ratios and Adversarial Contamination

This report synthesises findings from 15 peer-reviewed papers addressing the following research question: What is the impact of mixed human- and LLM-generated data ratios on the F1-score robustness of Llama3 models fine-tuned with different alignment techniques under increasing adversarial contamination.
openaire   +1 more source

Can Machine Learning Reduce Unnecessary Surgeries? A Retrospective Analysis Using Threshold Optimization to Prevent Negative Appendectomies in Adults

Annals of Gastroenterological Surgery, EarlyView.
Threshold‐optimized machine learning models using routine clinical and laboratory data in 623 adults undergoing appendectomy. Logistic regression (AUC = 0.765) and random forest (AUC = 0.785) were the best‐performing models for appendicitis detection and complicated appendicitis prediction, respectively.
Ivan Males, Marko Kumric, Zvonimir Boban, Josip Vrdoljak, Dario Pecenkovic, Matej Ivanda, Marko Grahovac, Zenon Pogorelic, Josko Bozic   +8 more
wiley   +1 more source

Novel Generative Metrics vs. F1-Scores for Robust Tabular Data Evaluation

This report synthesises findings from 10 peer-reviewed papers addressing the following research question: What is the comparative robustness of novel generative evaluation metrics versus traditional F1-scores when assessing model performance on progressively sparsified tabular datasets.
openaire   +1 more source

Harnessing Large Language Models to Advance Microbiome Research: From Sequence Analysis to Clinical Applications

Advanced Intelligent Discovery, EarlyView.
Large language models are transforming microbiome research by enabling advanced sequence profiling, functional prediction, and association mining across complex datasets. They automate microbial classification and disease‐state recognition, improving cross‐study integration and clinical diagnostics.
Jieqi Xing, He Wang, Yangyang Sun, Xiaoquan Su, Shunyao Wu   +4 more
wiley   +1 more source

Macrophage Phenotype Detection Methodology on Textured Surfaces via Nuclear Morphology Using Machine Learning

Advanced Intelligent Discovery, EarlyView.
A novel machine learning approach classifies macrophage phenotypes with up to 98% accuracy using only nuclear morphology from DAPI‐stained images. Bypassing traditional surface markers, the method proves robust even on complex textured biomaterial surfaces. It offers a simpler, faster alternative for studying macrophage behavior in various experimental
Oleh Mezhenskyi, Kateryna Andriienkova, Tetyana Goldanova, Ezgi Antmen, Thibaut Soubrie, Nihal Engin Vrana   +5 more
wiley   +1 more source

LRF-CNN: An explainable lightweight receptive field-based CNN for colorectal cancer histopathological image classification. [PDF]

iScience
Yuan L, Li C, Zhang J, Sun H, Rahaman M, Grzegorzek M, Li X.   +6 more
europepmc   +1 more source

Artificial Intelligence‐Driven Insights into Electrospinning: Machine Learning Models to Predict Cotton‐Wool‐Like Structure of Electrospun Fibers

Advanced Intelligent Discovery, EarlyView.
Electrospinning allows the fabrication of fibrous 3D cotton‐wool‐like scaffolds for tissue engineering. Optimizing this process traditionally relies on trial‐and‐error approaches, and artificial intelligence (AI)‐based tools can support it, with the prediction of fiber properties. This work uses machine learning to classify and predict the structure of
Paolo D’Elia, Giovanna Pappalardo, Aldo R. Boccaccini, Liliana Liverani   +3 more
wiley   +1 more source

Fault-tolerant pedigree reconstruction from pairwise kinship relations. [PDF]

Bioinformatics
Huang EC, Li KA, Narasimhan VM.
europepmc   +1 more source

A Comprehensive Assessment and Benchmark Study of Large Atomistic Foundation Models for Phonons

Advanced Intelligent Discovery, EarlyView.
We benchmark six large atomistic foundation models on 2429 crystalline materials for phonon transport properties. The rapid development of universal machine learning potentials (uMLPs) has enabled efficient, accurate predictions of diverse material properties across broad chemical spaces.
Md Zaibul Anam, Ogheneyoma Aghoghovbia, Mohammed Al‐Fahdi, Lingyu Kong, Victor Fung, Ming Hu   +5 more
wiley   +1 more source