Open Source Machine Learning Research Articles

Predicting CoVID-19 community mortality risk using machine learning and development of an online prognostic tool.

BackgroundThe recent pandemic of CoVID-19 has emerged as a threat to global health security. There are very few prognostic models on CoVID-19 using machine learning.ObjectivesTo predict mortality among confirmed CoVID-19 patients in South Korea using machine learning and deploy the best performing algorithm as an open-source online prediction tool for decision-making.Materials and MethodsMortality for confirmed CoVID-19 patients (n = 3,524) between January 20, 2020 and May 30, 2020 was predicted using five machine learning algorithms (logistic regression, support vector machine, K nearest neighbor, random forest and gradient boosting). The performance of the algorithms was compared, and the best performing algorithm was deployed as an online prediction tool.ResultsThe logistic regression algorithm was the best performer in terms of discrimination (area under ROC curve = 0.830), calibration (Matthews Correlation Coefficient = 0.433; Brier Score = 0.036) and. The best performing algorithm (logistic regression) was deployed as the online CoVID-19 Community Mortality Risk Prediction tool named CoCoMoRP (https://ashis-das.shinyapps.io/CoCoMoRP/).ConclusionsWe describe the development and deployment of an open-source machine learning tool to predict mortality risk among CoVID-19 confirmed patients using publicly available surveillance data. This tool can be utilized by potential stakeholders such as health providers and policymakers to triage patients at the community level in addition to other approaches.

Does size matter? The relationship between predictive power of single-subject morphometric networks to spatial scale and edge weight.

Network-level analysis based on anatomical, pairwise similarities (e.g., cortical thickness) has been gaining increasing attention recently. However, there has not been a systematic study of the impact of spatial scale and edge definitions on predictive performance, which is necessary to obtain a clear understanding of their relative performance. In this study, we present a histogram-based approach to construct subject-wise weighted networks that enable a principled comparison across different methods of network analysis. We design several weighted networks based on three large publicly available datasets and perform a robust evaluation of their predictive power under four levels of separability. An interesting insight generated is that changes in nodal size (spatial scale) have no significant impact on predictive power among the three classification experiments and two disease cohorts studied, i.e., mild cognitive impairment and Alzheimer's disease from ADNI, and Autism from the ABIDE dataset. We also release an open source python package called graynet to enable others to leverage the novel network feature extraction algorithms presented here. These techniques and toolbox can also be applied to other modalities due to their domain- and feature-agnostic nature) in diverse applications of connectivity research. In addition, the findings from the ADNI dataset are replicated in the AIBL dataset using an open source machine learning tool called neuropredict.

Improved Activity Recognition Combining Inertial Motion Sensors and Electroencephalogram Signals.

Human activity recognition and neural activity analysis are the basis for human computational neureoethology research dealing with the simultaneous analysis of behavioral ethogram descriptions and neural activity measurements. Wireless electroencephalography (EEG) and wireless inertial measurement units (IMU) allow the realization of experimental data recording with improved ecological validity where the subjects can be carrying out natural activities while data recording is minimally invasive. Specifically, we aim to show that EEG and IMU data fusion allows improved human activity recognition in a natural setting. We have defined an experimental protocol composed of natural sitting, standing and walking activities, and we have recruited subjects in two sites: in-house ([Formula: see text]) and out-house ([Formula: see text]) populations with different demographics. Experimental protocol data capture was carried out with validated commercial systems. Classifier model training and validation were carried out with scikit-learn open source machine learning python package. EEG features consist of the amplitude of the standard EEG frequency bands. Inertial features were the instantaneous position of the body tracked points after a moving average smoothing to remove noise. We carry out three validation processes: a 10-fold cross-validation process per experimental protocol repetition, (b) the inference of the ethograms, and (c) the transfer learning from each experimental protocol repetition to the remaining repetitions. The in-house accuracy results were lower and much more variable than the out-house sessions results. In general, random forest was the best performing classifier model. Best cross-validation results, ethogram accuracy, and transfer learning were achieved from the fusion of EEG and IMUs data. Transfer learning behaved poorly compared to classification on the same protocol repetition, but it has accuracy still greater than 0.75 on average for the out-house data sessions. Transfer leaning accuracy among repetitions of the same subject was above 0.88 on average. Ethogram prediction accuracy was above 0.96 on average. Therefore, we conclude that wireless EEG and IMUs allow for the definition of natural experimental designs with high ecological validity toward human computational neuroethology research. The fusion of both EEG and IMUs signals improves activity and ethogram recognition.

Machine Learning in High-Alert Medication Treatment: A Study on the Cardiovascular Drug

The safety of high-alert medication treatment is still a challenge all over the world. Approximately one-half of adverse drug events (ADEs) are related to high-alert medications, which motivates us to improve the predicament faced in clinical practice. The purpose of this study is to use machine-learning techniques to predict the risk of high-alert medication treatment. Taking the cardiovascular drug digoxin as an example, we collected the records of 513 patients who received the pertinent therapy during hospitalization at a tertiary medical center in Taiwan. Considering serum digoxin concentration (SDC) is the primary indicator for assessing the risk of digoxin therapy, patients with SDC being controlled at the recommended range before their discharge were defined as a low-risk population; otherwise, patients were defined as the high-risk population. Weka 3.9.4—an open source machine learning software—was adopted to develop binary classification models to predict the risk of digoxin therapy by a number of machine-learning techniques, including k-nearest neighbors (kNN), decision tree (C4.5), support vector machine (SVM), random forest (RF), artificial neural network (ANN) and logistic regression (LGR). The results showed that the performance of RF was the best, followed by C4.5 and ANN; the remaining classifiers performed poorly. This study confirmed that machine-learning techniques can yield favorable prediction effectiveness for high-alert medication treatment, thereby decreasing the risk of ADEs and improving medication safety.

Validation of PATHFx 2.0: An open-source tool for estimating survival in patients undergoing pathologic fracture fixation.

Treatment decisions in patients with metastatic bone disease rely on accurate survival estimation. We developed the original PATHFx models using expensive, proprietary software and now seek to provide a more cost-effective solution. Using open-source machine learning software to create PATHFx version 2.0, we asked whether PATHFx 2.0 could be created using open-source methods and externally validated in two unique patient populations. The training set of a well-characterized, database records of 189 patients and the bnlearn package within R Version 3.5.1 (R Foundation for Statistical Computing), was used to establish a series of Bayesian belief network models designed to predict survival at 1, 3, 6, 12, 18, and 24 months. Each was externally validated in both a Scandinavian (n = 815 patients) and a Japanese (n = 261 patients) data set. Brier scores and receiver operating characteristic curves to assessed discriminatory ability. Decision curve analysis (DCA) evaluated whether models should be used clinically. DCA showed that the model should be used clinically at all time points in the Scandinavian data set. For the 1-month time point, DCA of the Japanese data set suggested to expect better outcomes assuming all patients will survive greater than 1 month. Brier scores for each curve demonstrate that the models are accurate at each time point. Statement of Clinical Significance: we successfully transitioned to PATHFx 2.0 using open-source software and externally validated it in two unique patient populations, which can be used as a cost-effective option to guide surgical decisions in patients with metastatic bone disease.

Design and Development of Image Recognition Toolkit Based on Deep Learning

Deep learning algorithms have shown superior performance than traditional algorithms when dealing with computationally intensive tasks in many fields. The algorithm model based on deep learning has good performance and can improve the recognition accuracy in relevant applications in the field of computer vision. TensorFlow is a flexible opensource machine learning platform proposed by Google, which can run on a variety of platforms, such as CPU, GPU, and mobile devices. TensorFlow platform can also support current popular deep learning models. In this paper, an image recognition toolkit based on TensorFlow is designed and developed to simplify the development process of more and more image recognition applications. The toolkit uses convolutional neural networks to build a training model, which consists of two convolutional layers: one batch normalization layer before each convolutional layer, and the other pooling layer after each convolutional layer. The last two layers of the model use the full connection layer to output recognition results. Batch gradient descent algorithm is adopted in the optimization algorithm, and it integrates the advantages of both the gradient descent algorithm and the stochastic gradient descent algorithm, which greatly reduces the number of convergence iterations and has little influence on the convergence effect. The total training parameters of the toolkit model reach 1.7 million. In order to prevent overfitting problems, the dropout layer before each full connection layer is added and the threshold of 0.5 is set in the design. The convolution neural network model is trained and tested by the MNIST set on TensorFlow. The experimental result shows that the toolkit achieves the recognition accuracy of 99% on the MNIST test set. The development of the toolkit provides powerful technical support for the development of various image recognition applications, reduces its difficulty, and improves the efficiency of resource utilization.

GPU-Accelerated Vision for Robots: Improving System Throughput Using OpenCV and CUDA

OpenCV is an open source computer vision and machine learning library for C/C++/Python available for Windows, Linux, macOS, and Android platforms. It contains low-level image processing functions as well as high-level algorithms such as object identification, face recognition, and action classification in videos. OpenCV has become very popular, with more than 47,000 people in its user community and 18 million downloads (see https://opencv.org/about/). Under a Berkeley Software Distribution (BSD) license, it can be used for both academic and commercial applications.

A Blockchain and Machine Learning-Based Drug Supply Chain Management and Recommendation System for Smart Pharmaceutical Industry

From the last decade, pharmaceutical companies are facing difficulties in tracking their products during the supply chain process, allowing the counterfeiters to add their fake medicines into the market. Counterfeit drugs are analyzed as a very big challenge for the pharmaceutical industry worldwide. As indicated by the statistics, yearly business loss of around $200 billion is reported by US pharmaceutical companies due to these counterfeit drugs. These drugs may not help the patients to recover the disease but have many other dangerous side effects. According to the World Health Organization (WHO) survey report, in under-developed countries every 10th drug use by the consumers is counterfeit and has low quality. Hence, a system that can trace and track drug delivery at every phase is needed to solve the counterfeiting problem. The blockchain has the full potential to handle and track the supply chain process very efficiently. In this paper, we have proposed and implemented a novel blockchain and machine learning-based drug supply chain management and recommendation system (DSCMR). Our proposed system consists of two main modules: blockchain-based drug supply chain management and machine learning-based drug recommendation system for consumers. In the first module, the drug supply chain management system is deployed using Hyperledger fabrics which is capable of continuously monitor and track the drug delivery process in the smart pharmaceutical industry. On the other hand, the N-gram, LightGBM models are used in the machine learning module to recommend the top-rated or best medicines to the customers of the pharmaceutical industry. These models have trained on well known publicly available drug reviews dataset provided by the UCI: an open-source machine learning repository. Moreover, the machine learning module is integrated with this blockchain system with the help of the REST API. Finally, we also perform several tests to check the efficiency and usability of our proposed system.

Effect of automated TIL quantification in early-stage melanoma on accuracy of standard T staging using AJCC guidelines.

10076 Background: Patients diagnosed with early stage melanoma are at risk of recurrence and death. Adjuvant therapy decreases risk but incurs toxicity and expense. While tumor-infiltrating lymphocytes (TILs) improve prognosis, studies have shown conflicting results due, at least in part, to inter-observer variability. Thus, TILs are not included in standard American Joint Committee on Cancer (AJCC) staging. Here, we quantitatively analyze TILs in hematoxylin and eosin (H&E) melanoma images using two machine learning algorithms. Methods: H&E images were evaluated by two methods for patients with resectable stage I-III melanoma from Columbia (N = 81) and validated using samples from Geisinger and Moffitt (N = 128). For both methods, H&E images were manually annotated using open source software, QuPath, to specify tumor regions. For Method A, images were divided into patches and, for each patch, a probability was generated to detect lymphocytes. Patches above a set threshold were considered to be “TIL positive”. Ratio of TIL positive patches to total patches was assessed for every image. For Method B, a classifier was manually trained in QuPath and then applied on each image to determine the ratio of the areas of all immune cells to all tumor cells as previously published. Cutoff values to define high and low risk groups were established based on a test set and then validated in an independent cohort. Results: Both methods distinguished patients with visceral recurrence from those without for the Columbia training set (Method A p = .0015, Method B p = .043). Using Method A, Kaplan-Meier curve at the selected cutoff also correlated significantly with disease specific survival (DSS) for Columbia (p = .022) and was validated in the Geisinger/Moffitt (p = .046) cohort. Cox analysis using Method A showed that TIL status predicted DSS in the validation set (p = .047) and added significantly to depth and ulceration (HR = 3.43, CI: 1.047-11.257, p = .042). Conclusions: Both open source machine learning algorithms find significantly higher TILs in patients who do not develop metastasis. Notably, Method A may add to standard predictors, such as depth and ulceration. These results demonstrate the promise of computational algorithms to enhance visual grading, and suggest that digital TIL evaluation may add to current AJCC staging. [Table: see text]

Energy Demand Forecasting Using Deep Learning: Applications for the French Grid

This paper investigates the use of deep learning techniques in order to perform energy demand forecasting. To this end, the authors propose a mixed architecture consisting of a convolutional neural network (CNN) coupled with an artificial neural network (ANN), with the main objective of taking advantage of the virtues of both structures: the regression capabilities of the artificial neural network and the feature extraction capacities of the convolutional neural network. The proposed structure was trained and then used in a real setting to provide a French energy demand forecast using Action de Recherche Petite Echelle Grande Echelle (ARPEGE) forecasting weather data. The results show that this approach outperforms the reference Réseau de Transport d’Electricité (RTE, French transmission system operator) subscription-based service. Additionally, the proposed solution obtains the highest performance score when compared with other alternatives, including Autoregressive Integrated Moving Average (ARIMA) and traditional ANN models. This opens up the possibility of achieving high-accuracy forecasting using widely accessible deep learning techniques through open-source machine learning platforms.

Robust Intelligent Malware Detection using Light GBM Algorithm

Attackers take advantage of every second that the anti- vendor delays identifying the attacking malware signature and to provide notifications. In addition, the longer the detection period delayed, the greater the damage to the host device. To put it another way, the lack of ability to detect attacks early complicates the problem and rises serious harm. Consequently, this research intends to develop a knowledgeable anti-malware system capable of immediately detecting and terminating malware actions, rather than waiting for anti-malware updates. The research concentrates in its scope on the detection of malware on the Internet of Things (IoT), based on Machine Learning (ML) techniques. A latest open source ML algorithm called the Light Gradient Boosting Algorithm (LightGBM) has been used to develop our instant host and network layer anti-malware approach without any human intervention. For examination reasons, the suggested approach serves the LightGBM machine learning algorithm to adopt datasets obtained from real IoT devices using the LightGBM machine learning algorithm. The results indicate a successful method to detecting and classifying high accuracy malware at both network and host levels based on the Holdout method of cross-validation. Additionally, this result is better than many prior related studies which used different algorithms of Machine Learning and Deep Learning. Though, an old study which used the same dataset was the best among the literature. However, it still slightly less than what this study achieved, besides the complexity which deep learning adds. Lastly, the results show the ability of the proposed approach to detect IoT botnet attacks fast, which is a vital feature to end botnet activity before spreading to any new network device.

Data Pre Processing for Machine Learning Models using Python Libraries

Data pre-processing is the process of transforming the raw data into useful dataset. Data pre-processing is one of the most important phase of any machine learning model because the quality and efficiency of any machine learning model directly depends upon the data-set, if we skip this step and design a model with data sets containing missing values then the model we have designed will not be that efficient and will be inconsistent model. This paper describes the methodology for pre-processing the data in seven sequence of steps using python powerful libraries which are open source machine learning libraries that support both supervised and unsupervised learning like pandas is a high level data manipulation tool, scikit learn which provides various tools for model fitting, data pre-processing, model selection and many other utilities. These steps include dealing with missing value, categorical values, importing data sets etc. This analysis helps in cleaning and transforming the datasets which future applied to any learning model and produce a efficient machine learning model.

Abstract WP406: Training and Validation of Deepmedic Machine Learning Tool for Automated Hematoma Segmentation and Volume Analysis on CT Using Multicenter Data

Introduction: We sought to train and validate an automated machine learning algorithm for ICH segmentation and volume calculation using multicenter data. Methods: An open-source 3D deep machine learning algorithm “DeepMedic” was trained using manually segmented ICH from 208 CT scans (129 patients) from the multicenter PREDICT study. The algorithm was then validated with 125 manually segmented CT scans (48 patients) from the SPOTLIGHT study. Manual segmentation was performed with Quantomo semi-automated software. ABC/2 was measured for all studies by two neuroradiologists. Accuracy of DeepMedic segmentation was assessed using the Dice similarity coefficient. Analysis was stratified by presence of IVH. Intraclass correlation (ICC) with 95% confidence intervals (CI) assessed agreement between manual vs. DeepMedic segmentation volume; and manual segmentation and ABC/2 volume. Bland-Altman charts were analyzed for ABC/2 and DeepMedic vs. manual segmentation volumes. Results: DeepMedic demonstrated high segmentation accuracy in the training cohort (median Dice 0.96; IQR 0.95 - 0.97) and in the validation cohort (median Dice 0.91; IQR 0.86 - 0.94). Dice coefficients were not significantly different between patients with IVH in the training cohort; however was significantly worse in the validation cohort in patients with IVH (Wilcoxon p<0.001). Agreement was significantly better between DeepMedic and manual segmentation (PREDICT: ICC 0.99 [95%CI 0.99 -1.00]; SPOTLIGHT: ICC 0.98 [95%CI 0.97 - 0.99]) than between ABC/2 and manual segmentation (PREDICT: ICC 0.92 [95%CI 0.89 - 0.95]; SPOTLIGHT: ICC 0.95 [95%CI 0.93-0.97]). Improved accuracy of DeepMedic was demonstrated in Bland-Altman charts (Fig 1). Conclusion: ICH machine learning segmentation with DeepMedic is feasible and accurate; and demonstrates greater agreement with manual segmentation compared to ABC/2 volumes. Accuracy of the machine learning algorithm however is limited in patients with IVH.

Semi-Automated 3D Segmentation of Human Skeletal Muscle using Focused Ion Beam-Scanning Electron Microscopic Images Reveals Network of Mitochondria

Focused Ion Beam Scanning Electron Microscopy (FIB-SEM) is an imaging approach that enables analysis of the 3D architecture of cells and tissues at resolutions that are 1-2 orders of magnitude higher than that possible with light microscopy. The slow speeds of data collection and manual segmentation are two critical problems that limit the more extensive use of FIB-SEM technology. Our team developed an easily accessible robust method that enables rapid, large-scale acquisition of data from tissue specimens, combined with an approach for semi-automated data segmentation using the open-source machine learning Weka segmentation software, which dramatically increased the speed of image analysis.

Research and implementation of NN based on TensorFlow

TensorFlow is Google’s open-source machine learning and deep learning framework, which is convenient and flexible to build the current mainstream deep learning model. Convolutional neural network is a classical model of deep learning, the advantage lies in its powerful feature extraction capabilities of convolutional blocks. A neural network in the simplest case is a mathematical model consisting of several layers of elements that perform parallel calculations. Initially, such an architecture was created by analogy with the small computing elements of the human brain — neurons. The minimal computing elements of an artificial neural network are also called neurons. Neural networks typically consist of three or more layers: an input layer, a hidden layer (or layers), and an output layer. An important feature of the neural network is its ability to learn by example, this is called learning with a teacher. The neural network is trained on a large number of examples consisting of input-output pairs (corresponding to each other input and output). In object recognition problems, such a pair will be the input image and the corresponding label — the name of the object. Neural network learning is an iterative process that reduces the deviation of the network output from a given «teacher response» — a label that corresponds to a given image. This process consists of steps called epochs of learning (they are usually calculated in thousands), each of which is the adjustment of the «weights» of the neural network — the parameters of the hidden layers of the network. Upon completion of the learning process, the quality of the neural network is usually good enough to perform the task for which it was trained, although the optimal set of parameters that perfectly recognizes all the images, it is often impossible to choose. Based on the TensorFlow platform, a convolutional neural network model with two-convolution-layers was built. The model was trained and tested with the MNIST data set. The test accuracy rate could reach 99,15%, and compared with the rate of 98,69% with only one-convolution-layer model, which shows that the two-convolution-layers convolutional neural network model has a better ability of feature extraction and classification decision-making.

Overview on Open Source Machine Learning Platforms-TensorFlow

Deep learning is a part of artificial intelligence utilizing deep neural network architectures that have essentially progressed to the cutting edge in PC vision, speech recognition, characteristic language preparation, and different spaces. In November 2015, Google delivered TensorFlow, an open-source deep-learning programming library for characterizing, preparing, and conveying machine-learning models. In this paper, we audit Tensor Flow and put it in current deep learning ideas and programming. We discuss its essential computational standards and appropriated execution model, its programming interface, and visualization toolbox. We, at that point, contrast Tensor Flow with elective libraries, for example, Theano, Torch, or Caffe on a subjective just as a quantitative premise. Lastly, remark on watched use-instances of TensorFlow in the scholarly community and industry

Open Source Machine Learning Frameworks for Industrial Internet of Things

Abstract Information and communication technology has revolutionized the industrial operations and productions. The industries irrespective of size, whether small or large, have felt the need of artificial intelligence and machine learning techniques to process the terabytes of data generated through sensors, actuators, industrial management systems, and web applications. These data have the characteristics of volume (terabyte) and variety (image, audio, video, graphics) and thus customized models and techniques are required for analysis and management. The advancement in computer hardware, processing power, storage capacity, and cloud computing have led to experimentation and implementation of machine learning models in industrial domain for resource optimization, operation management, and quality control. However, the industrial Data Analysts face the dilemma of selecting the affordable and easy to use machine learning frameworks that suite their need and expectations. The study investigates the open source machine learning frameworks, aligned with the industrial domain (processing data generated from Industrial Internet of Things), in terms of usage, programming languages, implementations, and future prospects.

Deciphering tip-enhanced Raman imaging of carbon nanotubes with deep learning neural networks.

Recent release of open-source machine learning libraries presents opportunities to unify machine learning with nanoscale research, thus improving effectiveness of research methods and characterization protocols. This paper outlines and demonstrates the effectiveness of such a synergy with artificial neural networks to provide for an accelerated and enhanced characterization of individual carbon nanotubes deposited over a surface. Our algorithms provide a rapid diagnosis and analysis of tip-enhanced Raman spectroscopy mappings and the results show an improved spectral assignment of spectral features and spatial contrast of the collected images. Using several examples, we demonstrate the robustness and versatility of our deep learning neural network models. We highlight the use of machine learning and data science in tandem with tip-enhanced Raman spectroscopy technique enables a fast and accurate understanding of experimental data, thus leading to a powerful and comprehensive imaging analysis applied to spectroscopic measurements.

Efficiency comparison of Python and RapidMiner

Machine learning is an important technique that helps companies, institutes, and humans to improve the quality of decision making. As there are many different free tools on the Internet for machine learning, we need comparisons, benchmarks to provide help in the selection of the appropriate analysis technique. This paper aims at providing a comparative study of the two most powerful and open-source machine learning tools Python and RapidMiner by using most common supervised machine learning techniques Decision Tree, Support Vector Machines, and Neural Networks on data analytic of PIMA Indian Diabetes Dataset and Mushroom Classification Dataset. The results reveal that the usage of RapidMiner provides better performance in terms of both accuracy and execution time.

Applying Machine Learning to Nanoindentation Data of (Nano-) Enhanced Composites

Carbon fiber reinforced polymers (CFRPs) are continuously gaining attention in aerospace and space applications, and especially their multi-scale reinforcement with nanoadditives. Carbon nanotubes (CNTs), graphene, carbon nanofibers (CNFs), and their functionalized forms are often incorporated into interactive systems to engage specific changes in the environment of application to a smart response. Structural integrity of these nanoscale reinforced composites is assessed with advanced characterization techniques, with the most prominent being nanoindentation testing. Nanoindentation is a well-established technique, which enables quantitative mapping of nanomechanical properties with the μm surficial and nm indentation resolution scale and high precision characterization. This feature enables the characterization of the interface in a statistical and quantitative manner and the correlation of (nano-) reinforcement to interface properties of CFRPs. Identification of reinforcement is performed with k-Nearest Neighbors and Support Vector Machine classification algorithms. Expertise is necessary to describe the physical problem and create representative training/testing datasets. Development of open source Machine Learning algorithms can have an influential impact on uniformity of nanometry data creation and management. The statistical character of nanoindentation is a key factor to supply information on heterogeneity of multiscale reinforced composites. Both the identification of (nano-) reinforcement and quality assessment of composites are provided by involving artificial intelligence.