AI-Powered Microfluidics: Shaping the Future of Phenotypic Drug Discovery

ACS Applied Materials & Interfaces, Journal Year: 2024, Volume and Issue: 16(30), P. 38832 - 38851

Published: July 17, 2024

Phenotypic drug discovery (PDD), which involves harnessing biological systems directly to uncover effective drugs, has undergone a resurgence in recent years. The rapid advancement of artificial intelligence (AI) over the past few years presents numerous opportunities for augmenting phenotypic screening on microfluidic platforms, leveraging its predictive capabilities, data analysis, efficient processing, etc. Microfluidics coupled with AI is poised revolutionize landscape discovery. By integrating advanced platforms algorithms, researchers can rapidly screen large libraries compounds, identify novel candidates, and elucidate complex pathways unprecedented speed efficiency. This review provides an overview advances challenges AI-based microfluidics their applications We discuss synergistic combination high-throughput AI-driven analysis phenotype characterization, drug-target interactions, modeling. In addition, we highlight potential AI-powered achieve automated system. Overall, represents promising approach shaping future by enabling rapid, cost-effective, accurate identification therapeutically relevant compounds.

Language: Английский

---

BioDeepfuse: a hybrid deep learning approach with integrated feature extraction techniques for enhanced non-coding RNA classification

RNA Biology, Journal Year: 2024, Volume and Issue: 21(1), P. 1 - 12

Published: March 25, 2024

The accurate classification of non-coding RNA (ncRNA) sequences is pivotal for advanced genome annotation and analysis, a fundamental aspect genomics that facilitates understanding ncRNA functions regulatory mechanisms in various biological processes. While traditional machine learning approaches have been employed distinguishing ncRNA, these often necessitate extensive feature engineering. Recently, deep algorithms provided advancements classification. This study presents BioDeepFuse, hybrid framework integrating convolutional neural networks (CNN) or bidirectional long short-term memory (BiLSTM) with handcrafted features enhanced accuracy. employs combination k-mer one-hot, dictionary, extraction techniques input representation. Extracted features, when embedded into the network, enable optimal utilization spatial sequential nuances sequences. Using benchmark datasets real-world samples from bacterial organisms, we evaluated performance BioDeepFuse. Results exhibited high accuracy classification, underscoring robustness our tool addressing complex sequence data challenges. effective melding CNN BiLSTM external heralds promising directions future research, particularly refining classifiers deepening insights ncRNAs cellular processes disease manifestations. In addition to its original application context methodologies integrated can potentially render BioDeepFuse broader domains.

Language: Английский

---

A novel extended multimodal AI framework towards vulnerability detection in smart contracts

Information Sciences, Journal Year: 2023, Volume and Issue: 636, P. 118907 - 118907

Published: March 23, 2023

Language: Английский

---

An Effective WBC Segmentation and Classification Using MobilenetV3–ShufflenetV2 Based Deep Learning Framework

IEEE Access, Journal Year: 2023, Volume and Issue: 11, P. 27739 - 27748

Published: Jan. 1, 2023

White Blood Cells are essential in keeping track of a person's health. However, the pathologist's experience will determine how blood smear is evaluated. Furthermore, it still challenging to classify WBCs accurately because they have various forms, sizes, and colors due distinct cell subtypes labeling methods. As result, powerful deep learning system for WBC categorization based on MobilenetV3-ShufflenetV2 described this research. Initially, images segmented using an efficient Pyramid Scene Parsing Network (PSPNet). Following that, MobilenetV3 Artificial Gravitational Cuckoo Search (AGCS)-based technique used extract select global local features from images. Finally, divided into five classes ShufflenetV2 model. The proposed approach evaluated count detection (BCCD) Raabin-Wbc datasets achieves 99.19% 99% accuracy, respectively. Moreover, results satisfactory when compared existing algorithms.

Language: Английский

---

A review of convolutional neural network based methods for medical image classification

Computers in Biology and Medicine, Journal Year: 2024, Volume and Issue: 185, P. 109507 - 109507

Published: Dec. 3, 2024

Language: Английский

---

EfficientNet - XGBoost: An Effective White-Blood-Cell Segmentation and Classification Framework

Nano Biomedicine and Engineering, Journal Year: 2023, Volume and Issue: 15(2), P. 126 - 135

Published: May 26, 2023

In the human body, white blood cells (WBCs) are crucial immune that help in early detection of a variety illnesses. Determination number WBCs can be used to diagnose conditions such as hematological, immunological, and autoimmune diseases, well AIDS leukemia. However, conventional method classifying counting is time-consuming, laborious, potentially erroneous. Therefore, this paper presents computer-assisted automated for recognizing detecting WBC categories from images. Initially, cell image preprocessed then segmented using an effective deep learning architecture called SegNet. Then, important features devised extracted EfficientNet architecture. Finally, categorized into four different types XGBoost classifier: neutrophils, eosinophils, monocytes, lymphocytes. The advantages SegNet, EfficientNet, make proposed model more robust achieve efficient classification WBCs. BCCD dataset evaluate performance methodology, findings compared existing state-of-the-art approaches based on accuracy, precision, sensitivity, specificity, F1-score. Evaluation results show approach has higher rank-1 accuracy 99.02% outperformed other techniques.

Language: Английский

---

An Image Classification Method of Unbalanced Ship Coating Defects Based on DCCVAE-ACWGAN-GP

Coatings, Journal Year: 2024, Volume and Issue: 14(3), P. 288 - 288

Published: Feb. 27, 2024

Affected by the improper operation of workers, environmental changes during drying and curing or quality paint itself, diverse defects are produced process ship painting. The traditional defect recognition method relies on expert knowledge experience to detect defects, which is not conducive ensuring effectiveness recognition. Therefore, this paper proposes an image generation model suitable for small samples. Based a deep convolutional neural network (DCNN), combines conditional variational autoencoder (DCCVAE) auxiliary Wasserstein GAN with gradient penalty (ACWGAN-GP) gradually expand generate various coating images solving overfitting problem due unbalanced data. DCNN trained based newly generated data original so as build classification samples, improving performance. experimental results showed that our proposed can achieve up 92.54% accuracy, F-score 88.33%, G mean value 91.93%. Compared enhancement methods algorithms, identify in painting more accurately consistently, provide effective theoretical technical support detection has significant engineering research application prospects.

Language: Английский

---

Madhubani Art Classification using transfer learning with deep feature fusion and decision fusion based techniques

Engineering Applications of Artificial Intelligence, Journal Year: 2022, Volume and Issue: 119, P. 105734 - 105734

Published: Dec. 21, 2022

Language: Английский

---

iTCep: a deep learning framework for identification of T cell epitopes by harnessing fusion features

Frontiers in Genetics, Journal Year: 2023, Volume and Issue: 14

Published: May 9, 2023

Neoantigens recognized by cytotoxic T cells are effective targets for tumor-specific immune responses personalized cancer immunotherapy. Quite a few neoantigen identification pipelines and computational strategies have been developed to improve the accuracy of peptide selection process. However, these methods mainly consider end ignore interaction between peptide-TCR preference each residue in TCRs, resulting filtered peptides often fail truly elicit an response. Here, we propose novel encoding approach representation. Subsequently, deep learning framework, namely iTCep, was predict interactions TCRs using fusion features derived from feature-level strategy. The iTCep achieved high predictive performance with AUC up 0.96 on testing dataset above 0.86 independent datasets, presenting better prediction compared other predictors. Our results provided strong evidence that model can be reliable robust method predicting TCR binding specificities given antigen peptides. One access through user-friendly web server at http://biostatistics.online/iTCep/ , which supports modes pairs peptide-only. A stand-alone software program cell epitope is also available convenient installing https://github.com/kbvstmd/iTCep/ .

Language: Английский

---

Enhancing Disease Diagnosis: A CNN-Based Approach for Automated White Blood Cell Classification

2021 IEEE International Conference on Big Data (Big Data), Journal Year: 2023, Volume and Issue: unknown, P. 4606 - 4613

Published: Dec. 15, 2023

White Blood Cell (WBC) image classification is pivotal for early disease detection and diagnosis. Convolutional Neural Networks (CNNs) have emerged as potent tools such tasks due to their ability learn intricate features from raw pixel data. In this study, we present a CNN-based approach automated WBC classification. Our methodology encompasses preprocessing enhance contrast normalize color, succeeded by CNN training with multiple convolutional pooling layers, thereby enabling feature acquisition diverse classes. We evaluate our using publicly accessible dataset, comparing results against other contemporary methods. proposed method achieves an impressive 96.2% accuracy six distinct classes, surpassing prior techniques considerable margin. This showcases CNNs' potential in classification, underscoring its significance medical diagnosis research. summary, introduce that attains state-of-the-art performance on available dataset. preprocessing, enhancement, color normalization, capture distinctive of findings underscore promise domain propose deployment valuable tool research Subsequent efforts will explore advanced like transfer learning further elevate method's performance.

Language: Английский

---