A guide to deep learning in healthcare

Nature Medicine, Journal Year: 2018, Volume and Issue: 25(1), P. 24 - 29

Published: Dec. 24, 2018

Language: Английский

---

CellProfiler 3.0: Next-generation image processing for biology

PLoS Biology, Journal Year: 2018, Volume and Issue: 16(7), P. e2005970 - e2005970

Published: July 3, 2018

CellProfiler has enabled the scientific research community to create flexible, modular image analysis pipelines since its release in 2005. Here, we describe 3.0, a new version of software supporting both whole-volume and plane-wise three-dimensional (3D) stacks, increasingly common biomedical research. CellProfiler's infrastructure is greatly improved, provide protocol for cloud-based, large-scale processing. New plugins enable running pretrained deep learning models on images. Designed by biologists, equips researchers with powerful computational tools via well-documented user interface, empowering biologists all fields quantitative, reproducible workflows.

Language: Английский

---

The rise of deep learning in drug discovery

Drug Discovery Today, Journal Year: 2018, Volume and Issue: 23(6), P. 1241 - 1250

Published: Jan. 31, 2018

Over the past decade, deep learning has achieved remarkable success in various artificial intelligence research areas. Evolved from previous on neural networks, this technology shown superior performance to other machine algorithms areas such as image and voice recognition, natural language processing, among others. The first wave of applications pharmaceutical emerged recent years, its utility gone beyond bioactivity predictions promise addressing diverse problems drug discovery. Examples will be discussed covering prediction, de novo molecular design, synthesis prediction biological analysis.

Language: Английский

---

Recent applications of deep learning and machine intelligence on in silico drug discovery: methods, tools and databases

Briefings in Bioinformatics, Journal Year: 2018, Volume and Issue: 20(5), P. 1878 - 1912

Published: June 16, 2018

The identification of interactions between drugs/compounds and their targets is crucial for the development new drugs. In vitro screening experiments (i.e. bioassays) are frequently used this purpose; however, experimental approaches insufficient to explore novel drug-target interactions, mainly because feasibility problems, as they labour intensive, costly time consuming. A computational field known 'virtual screening' (VS) has emerged in past decades aid drug discovery studies by statistically estimating unknown bio-interactions compounds biological targets. These methods use physico-chemical structural properties and/or target proteins along with experimentally verified bio-interaction information generate predictive models. Lately, sophisticated machine learning techniques applied VS elevate performance. objective study examine discuss recent applications VS, including deep learning, which became highly popular after giving rise epochal developments fields computer vision natural language processing. 3 years have witnessed an unprecedented amount research considering application biomedicine, discovery. review, we first describe main instruments methods, compound protein features representations descriptors), libraries toolkits bioactivity databases gold-standard data sets system training benchmarking. We subsequently review a strong emphasis on applications. Finally, present state field, current challenges suggest future directions. believe that survey will provide insight researchers working terms comprehending developing bio-prediction methods.

Language: Английский

---

Breast cancer detection using deep convolutional neural networks and support vector machines

PeerJ, Journal Year: 2019, Volume and Issue: 7, P. e6201 - e6201

Published: Jan. 28, 2019

It is important to detect breast cancer as early possible. In this manuscript, a new methodology for classifying using deep learning and some segmentation techniques are introduced. A computer aided detection (CAD) system proposed benign malignant mass tumors in mammography images. CAD system, two approaches used. The first approach involves determining the region of interest (ROI) manually, while second uses technique threshold based. convolutional neural network (DCNN) used feature extraction. well-known DCNN architecture named AlexNet fine-tuned classify classes instead 1,000 classes. last fully connected (fc) layer support vector machine (SVM) classifier obtain better accuracy. results obtained following publicly available datasets (1) digital database screening (DDSM); (2) Curated Breast Imaging Subset DDSM (CBIS-DDSM). Training on large number data gives high accuracy rate. Nevertheless, biomedical contain relatively small samples due limited patient volume. Accordingly, augmentation method increasing size input by generating from original data. There many forms augmentation; one here rotation. new-trained 71.01% when cropping ROI manually mammogram. highest area under curve (AUC) achieved was 0.88 (88%) both techniques. Moreover, CBIS-DDSM, increased 73.6%. Consequently, SVM becomes 87.2% with an AUC equaling 0.94 (94%). This value compared previous work same conditions.

Language: Английский

---

Deep Learning in Mining Biological Data

Cognitive Computation, Journal Year: 2021, Volume and Issue: 13(1), P. 1 - 33

Published: Jan. 1, 2021

Recent technological advancements in data acquisition tools allowed life scientists to acquire multimodal from different biological application domains. Categorized three broad types (i.e. images, signals, and sequences), these are huge amount complex nature. Mining such enormous of for pattern recognition is a big challenge requires sophisticated data-intensive machine learning techniques. Artificial neural network-based systems well known their capabilities, lately deep architectures-known as (DL)-have been successfully applied solve many problems. To investigate how DL-especially its architectures-has contributed utilized the mining pertaining those types, meta-analysis has performed resulting resources have critically analysed. Focusing on use DL analyse patterns diverse domains, this work investigates architectures' applications data. This followed by an exploration available open access sources along with popular open-source applicable Also, comparative investigations qualitative, quantitative, benchmarking perspectives provided. Finally, some research challenges using mine outlined number possible future put forward.

Language: Английский

---

A clinical text classification paradigm using weak supervision and deep representation

BMC Medical Informatics and Decision Making, Journal Year: 2019, Volume and Issue: 19(1)

Published: Jan. 7, 2019

Automatic clinical text classification is a natural language processing (NLP) technology that unlocks information embedded in narratives. Machine learning approaches have been shown to be effective for tasks. However, successful machine model usually requires extensive human efforts create labeled training data and conduct feature engineering. In this study, we propose paradigm using weak supervision deep representation reduce these efforts.

Language: Английский

---

LSTMVis: A Tool for Visual Analysis of Hidden State Dynamics in Recurrent Neural Networks

IEEE Transactions on Visualization and Computer Graphics, Journal Year: 2017, Volume and Issue: 24(1), P. 667 - 676

Published: Aug. 28, 2017

Recurrent neural networks, and in particular long short-term memory (LSTM) are a remarkably effective tool for sequence modeling that learn dense black-box hidden representation of their sequential input. Researchers interested better understanding these models have studied the changes state representations over time noticed some interpretable patterns but also significant noise. In this work, we present LSTMVis, visual analysis recurrent networks with focus on dynamics. The allows users to select hypothesis input range local changes, match states similar large data set, align results structural annotations from domain. We show several use cases analyzing specific properties dataset containing nesting, phrase structure, chord progressions, demonstrate how can be used isolate further statistical analysis. characterize domain, different stakeholders, goals tasks. Long-term usage after putting online revealed great interest machine learning community.

Language: Английский

---

Cox-nnet: An artificial neural network method for prognosis prediction of high-throughput omics data

PLoS Computational Biology, Journal Year: 2018, Volume and Issue: 14(4), P. e1006076 - e1006076

Published: April 10, 2018

Artificial neural networks (ANN) are computing architectures with many interconnections of simple neural-inspired elements, and have been applied to biomedical fields such as imaging analysis diagnosis. We developed a new ANN framework called Cox-nnet predict patient prognosis from high throughput transcriptomics data. In 10 TCGA RNA-Seq data sets, achieves the same or better predictive accuracy compared other methods, including Cox-proportional hazards regression (with LASSO, ridge, mimimax concave penalty), Random Forests Survival CoxBoost. also reveals richer biological information, at both pathway gene levels. The outputs hidden layer node provide an alternative approach for survival-sensitive dimension reduction. summary, we method accurate efficient prediction on data, functional insights. source code is freely available https://github.com/lanagarmire/cox-nnet.

Language: Английский

---

Deep autoencoder based energy method for the bending, vibration, and buckling analysis of Kirchhoff plates with transfer learning

European Journal of Mechanics - A/Solids, Journal Year: 2021, Volume and Issue: 87, P. 104225 - 104225

Published: Jan. 26, 2021

Language: Английский

---