Accurately Differentiating Between Patients With COVID-19, Patients With Other Viral Infections, and Healthy Individuals: Multimodal Late Fusion Learning Approach

Background Effectively identifying patients with COVID-19 using nonpolymerase chain reaction biomedical data is critical for achieving optimal clinical outcomes. Currently, there is a lack of comprehensive understanding in various biomedical features and appropriate analytical approaches for enabling the early detection and effective diagnosis of patients with COVID-19. Objective We aimed to combine low-dimensional clinical and lab testing data, as well as high-dimensional computed tomography (CT) imaging data, to accurately differentiate between healthy individuals, patients with COVID-19, and patients with non-COVID viral pneumonia, especially at the early stage of infection. Methods In this study, we recruited 214 patients with nonsevere COVID-19, 148 patients with severe COVID-19, 198 noninfected healthy participants, and 129 patients with non-COVID viral pneumonia. The participants’ clinical information (ie, 23 features), lab testing results (ie, 10 features), and CT scans upon admission were acquired and used as 3 input feature modalities. To enable the late fusion of multimodal features, we constructed a deep learning model to extract a 10-feature high-level representation of CT scans. We then developed 3 machine learning models (ie, k-nearest neighbor, random forest, and support vector machine models) based on the combined 43 features from all 3 modalities to differentiate between the following 4 classes: nonsevere, severe, healthy, and viral pneumonia. Results Multimodal features provided substantial performance gain from the use of any single feature modality. All 3 machine learning models had high overall prediction accuracy (95.4%-97.7%) and high class-specific prediction accuracy (90.6%-99.9%). Conclusions Compared to the existing binary classification benchmarks that are often focused on single-feature modality, this study’s hybrid deep learning-machine learning framework provided a novel and effective breakthrough for clinical applications. Our findings, which come from a relatively large sample size, and analytical workflow will supplement and assist with clinical decision support for current COVID-19 diagnostic methods and other clinical applications with high-dimensional multimodal biomedical features.

Download Full-text

Accurately Differentiating Between Patients With COVID-19, Patients With Other Viral Infections, and Healthy Individuals: Multimodal Late Fusion Learning Approach (Preprint)

10.2196/preprints.25535 ◽

2020 ◽

Author(s):

Ming Xu ◽

Liu Ouyang ◽

Lei Han ◽

Kai Sun ◽

Tingting Yu ◽

...

Keyword(s):

Machine Learning ◽

Prediction Accuracy ◽

Clinical Applications ◽

Viral Pneumonia ◽

High Dimensional ◽

Healthy Individuals ◽

Late Fusion ◽

Single Feature ◽

Multimodal Features ◽

Machine Learning Models

BACKGROUND Effectively identifying patients with COVID-19 using nonpolymerase chain reaction biomedical data is critical for achieving optimal clinical outcomes. Currently, there is a lack of comprehensive understanding in various biomedical features and appropriate analytical approaches for enabling the early detection and effective diagnosis of patients with COVID-19. OBJECTIVE We aimed to combine low-dimensional clinical and lab testing data, as well as high-dimensional computed tomography (CT) imaging data, to accurately differentiate between healthy individuals, patients with COVID-19, and patients with non-COVID viral pneumonia, especially at the early stage of infection. METHODS In this study, we recruited 214 patients with nonsevere COVID-19, 148 patients with severe COVID-19, 198 noninfected healthy participants, and 129 patients with non-COVID viral pneumonia. The participants’ clinical information (ie, 23 features), lab testing results (ie, 10 features), and CT scans upon admission were acquired and used as 3 input feature modalities. To enable the late fusion of multimodal features, we constructed a deep learning model to extract a 10-feature high-level representation of CT scans. We then developed 3 machine learning models (ie, k-nearest neighbor, random forest, and support vector machine models) based on the combined 43 features from all 3 modalities to differentiate between the following 4 classes: nonsevere, severe, healthy, and viral pneumonia. RESULTS Multimodal features provided substantial performance gain from the use of any single feature modality. All 3 machine learning models had high overall prediction accuracy (95.4%-97.7%) and high class-specific prediction accuracy (90.6%-99.9%). CONCLUSIONS Compared to the existing binary classification benchmarks that are often focused on single-feature modality, this study’s hybrid deep learning-machine learning framework provided a novel and effective breakthrough for clinical applications. Our findings, which come from a relatively large sample size, and analytical workflow will supplement and assist with clinical decision support for current COVID-19 diagnostic methods and other clinical applications with high-dimensional multimodal biomedical features.

Download Full-text

Accurately Differentiating COVID-19, Other Viral Infection, and Healthy Individuals Using Multimodal Features via Late Fusion Learning

10.1101/2020.08.18.20176776 ◽

2020 ◽

Author(s):

Ming Xu ◽

Liu Ouyang ◽

Yan Gao ◽

Yuanfang Chen ◽

Tingting Yu ◽

...

Keyword(s):

Nearest Neighbor ◽

Binary Classification ◽

Clinical Decision ◽

Clinical Applications ◽

Ct Scans ◽

Support Vector ◽

Late Fusion ◽

K Nearest Neighbor ◽

Single Feature ◽

Multimodal Features

Effectively identifying COVID-19 patients using non-PCR clinical data is critical for the optimal clinical outcomes. Currently, there is a lack of comprehensive understanding of various biomedical features and appropriate technical approaches to accurately detecting COVID-19 patients. In this study, we recruited 214 confirmed COVID-19 patients in non-severe (NS) and 148 in severe (S) clinical type, 198 non-infected healthy (H) participants and 129 non-COVID viral pneumonia (V) patients. The participants' clinical information (23 features), lab testing results (10 features), and thoracic CT scans upon admission were acquired as three input feature modalities. To enable late fusion of multimodality data, we developed a deep learning model to extract a 10-feature high-level representation of the CT scans. Exploratory analyses showed substantial differences of all features among the four classes. Three machine learning models (k-nearest neighbor kNN, random forest RF, and support vector machine SVM) were developed based on the 43 features combined from all three modalities to differentiate four classes (NS, S, V, and H) at once. All three models had high accuracy to differentiate the overall four classes (95.4%-97.7%) and each individual class (90.6%-99.9%). Multimodal features provided substantial performance gain from using any single feature modality. Compared to existing binary classification benchmarks often focusing on single feature modality, this study provided a novel and effective breakthrough for clinical applications. Findings and the analytical workflow can be used as clinical decision support for current COVID-19 and other clinical applications with high-dimensional multimodal biomedical features.

Download Full-text

Application of Various Machine Learning Models for Process Stability of Bio-Electrochemical Anaerobic Digestion

Processes ◽

10.3390/pr10010158 ◽

2022 ◽

Vol 10 (1) ◽

pp. 158

Author(s):

Ain Cheon ◽

Jwakyung Sung ◽

Hangbae Jun ◽

Heewon Jang ◽

Minji Kim ◽

...

Keyword(s):

Machine Learning ◽

Anaerobic Digestion ◽

Prediction Accuracy ◽

Input Data ◽

Process Stability ◽

Operational Parameters ◽

Monitoring Parameter ◽

Efficient Prediction ◽

Single Input ◽

Machine Learning Models

The application of a machine learning (ML) model to bio-electrochemical anaerobic digestion (BEAD) is a future-oriented approach for improving process stability by predicting performances that have nonlinear relationships with various operational parameters. Five ML models, which included tree-, regression-, and neural network-based algorithms, were applied to predict the methane yield in BEAD reactor. The results showed that various 1-step ahead ML models, which utilized prior data of BEAD performances, could enhance prediction accuracy. In addition, 1-step ahead with retraining algorithm could improve prediction accuracy by 37.3% compared with the conventional multi-step ahead algorithm. The improvement was particularly noteworthy in tree- and regression-based ML models. Moreover, 1-step ahead with retraining algorithm showed high potential of achieving efficient prediction using pH as a single input data, which is plausibly an easier monitoring parameter compared with the other parameters required in bioprocess models.

Download Full-text

The Role of Textualisation and Argumentation in Understanding the Machine Learning Process

Proceedings of the Twenty-Sixth International Joint Conference on Artificial Intelligence ◽

10.24963/ijcai.2017/765 ◽

2017 ◽

Author(s):

Kacper Sokol ◽

Peter Flach

Keyword(s):

Machine Learning ◽

Predictive Accuracy ◽

Spatial Perception ◽

Black Box ◽

High Dimensional ◽

Box Models ◽

Machine Learning Applications ◽

Black Box Models ◽

Machine Learning Models

Understanding data, models and predictions is important for machine learning applications. Due to the limitations of our spatial perception and intuition, analysing high-dimensional data is inherently difficult. Furthermore, black-box models achieving high predictive accuracy are widely used, yet the logic behind their predictions is often opaque. Use of textualisation -- a natural language narrative of selected phenomena -- can tackle these shortcomings. When extended with argumentation theory we could envisage machine learning models and predictions arguing persuasively for their choices.

Download Full-text

Online learning behavior analysis based on machine learning

Asian Association of Open Universities Journal ◽

10.1108/aaouj-08-2019-0029 ◽

2019 ◽

Vol 14 (2) ◽

pp. 97-106

Author(s):

Ning Yan ◽

Oliver Tat-Sheung Au

Keyword(s):

Machine Learning ◽

Online Learning ◽

Correlation Analysis ◽

Prediction Accuracy ◽

Classification Models ◽

Limited Data ◽

Learning Models ◽

Learning Behavior ◽

Content Type ◽

Machine Learning Models

Purpose The purpose of this paper is to make a correlation analysis between students’ online learning behavior features and course grade, and to attempt to build some effective prediction model based on limited data. Design/methodology/approach The prediction label in this paper is the course grade of students, and the eigenvalues available are student age, student gender, connection time, hits count and days of access. The machine learning model used in this paper is the classical three-layer feedforward neural networks, and the scaled conjugate gradient algorithm is adopted. Pearson correlation analysis method is used to find the relationships between course grade and the student eigenvalues. Findings Days of access has the highest correlation with course grade, followed by hits count, and connection time is less relevant to students’ course grade. Student age and gender have the lowest correlation with course grade. Binary classification models have much higher prediction accuracy than multi-class classification models. Data normalization and data discretization can effectively improve the prediction accuracy of machine learning models, such as ANN model in this paper. Originality/value This paper may help teachers to find some clue to identify students with learning difficulties in advance and give timely help through the online learning behavior data. It shows that acceptable prediction models based on machine learning can be built using a small and limited data set. However, introducing external data into machine learning models to improve its prediction accuracy is still a valuable and hard issue.

Download Full-text

Machine Learning for Localizing Epileptogenic-Zone in the Temporal Lobe: Quantifying the Value of Multimodal Clinical-Semiology and Imaging Concordance

Frontiers in Digital Health ◽

10.3389/fdgth.2021.559103 ◽

2021 ◽

Vol 3 ◽

Author(s):

Ali Alim-Marvasti ◽

Fernando Pérez-García ◽

Karan Dahele ◽

Gloria Romagnoli ◽

Beate Diehl ◽

...

Keyword(s):

Machine Learning ◽

Temporal Lobe ◽

Support Vector ◽

Seizure Freedom ◽

Epileptogenic Zone ◽

Learning Models ◽

Seizure Semiology ◽

Multimodal Features ◽

Machine Learning Models ◽

Better Than

Background: Epilepsy affects 50 million people worldwide and a third are refractory to medication. If a discrete cerebral focus or network can be identified, neurosurgical resection can be curative. Most excisions are in the temporal-lobe, and are more likely to result in seizure-freedom than extra-temporal resections. However, less than half of patients undergoing surgery become entirely seizure-free. Localizing the epileptogenic-zone and individualized outcome predictions are difficult, requiring detailed evaluations at specialist centers.Methods: We used bespoke natural language processing to text-mine 3,800 electronic health records, from 309 epilepsy surgery patients, evaluated over a decade, of whom 126 remained entirely seizure-free. We investigated the diagnostic performances of machine learning models using set-of-semiology (SoS) with and without hippocampal sclerosis (HS) on MRI as features, using STARD criteria.Findings: Support Vector Classifiers (SVC) and Gradient Boosted (GB) decision trees were the best performing algorithms for temporal-lobe epileptogenic zone localization (cross-validated Matthews correlation coefficient (MCC) SVC 0.73 ± 0.25, balanced accuracy 0.81 ± 0.14, AUC 0.95 ± 0.05). Models that only used seizure semiology were not always better than internal benchmarks. The combination of multimodal features, however, enhanced performance metrics including MCC and normalized mutual information (NMI) compared to either alone (p < 0.0001). This combination of semiology and HS on MRI increased both cross-validated MCC and NMI by over 25% (NMI, SVC SoS: 0.35 ± 0.28 vs. SVC SoS+HS: 0.61 ± 0.27).Interpretation: Machine learning models using only the set of seizure semiology (SoS) cannot unequivocally perform better than benchmarks in temporal epileptogenic-zone localization. However, the combination of SoS with an imaging feature (HS) enhance epileptogenic lobe localization. We quantified this added NMI value to be 25% in absolute terms. Despite good performance in localization, no model was able to predict seizure-freedom better than benchmarks. The methods used are widely applicable, and the performance enhancements by combining other clinical, imaging and neurophysiological features could be similarly quantified. Multicenter studies are required to confirm generalizability.Funding: Wellcome/EPSRC Center for Interventional and Surgical Sciences (WEISS) (203145Z/16/Z).

Download Full-text

An Evaluation of Wearable Inertial Sensor Configuration and Supervised Machine Learning Models for Automatic Punch Classification in Boxing

IoT ◽

10.3390/iot1020021 ◽

2020 ◽

Vol 1 (2) ◽

pp. 360-381

Author(s):

Matthew T. O. Worsey ◽

Hugo G. Espinosa ◽

Jonathan B. Shepherd ◽

David V. Thiel

Keyword(s):

Machine Learning ◽

Prediction Accuracy ◽

Inertial Sensors ◽

Inertial Sensor ◽

Supervised Machine Learning ◽

Learning Models ◽

Significant Statistical Difference ◽

Sensor Configuration ◽

Wearable Inertial Sensors ◽

Machine Learning Models

Machine learning is a powerful tool for data classification and has been used to classify movement data recorded by wearable inertial sensors in general living and sports. Inertial sensors can provide valuable biofeedback in combat sports such as boxing; however, the use of such technology has not had a global uptake. If simple inertial sensor configurations can be used to automatically classify strike type, then cumbersome tasks such as video labelling can be bypassed and the foundation for automated workload monitoring of combat sport athletes is set. This investigation evaluates the classification performance of six different supervised machine learning models (tuned and untuned) when using two simple inertial sensor configurations (configuration 1—inertial sensor worn on both wrists; configuration 2—inertial sensor worn on both wrists and third thoracic vertebrae [T3]). When trained on one athlete, strike prediction accuracy was good using both configurations (sensor configuration 1 mean overall accuracy: 0.90 ± 0.12; sensor configuration 2 mean overall accuracy: 0.87 ± 0.09). There was no significant statistical difference in prediction accuracy between both configurations and tuned and untuned models (p > 0.05). Moreover, there was no significant statistical difference in computational training time for tuned and untuned models (p > 0.05). For sensor configuration 1, a support vector machine (SVM) model with a Gaussian rbf kernel performed the best (accuracy = 0.96), for sensor configuration 2, a multi-layered perceptron neural network (MLP-NN) model performed the best (accuracy = 0.98). Wearable inertial sensors can be used to accurately classify strike-type in boxing pad work, this means that cumbersome tasks such as video and notational analysis can be bypassed. Additionally, automated workload and performance monitoring of athletes throughout training camp is possible. Future investigations will evaluate the performance of this algorithm on a greater sample size and test the influence of impact window-size on prediction accuracy. Additionally, supervised machine learning models should be trained on data collected during sparring to see if high accuracy holds in a competition setting. This can help move closer towards automatic scoring in boxing.

Download Full-text

A Dynamic Convolutional Neural Network Based Shared-Bike Demand Forecasting Model

ACM Transactions on Intelligent Systems and Technology ◽

10.1145/3447988 ◽

2021 ◽

Vol 12 (6) ◽

pp. 1-24

Author(s):

Shaojie Qiao ◽

Nan Han ◽

Jianbin Huang ◽

Kun Yue ◽

Rui Mao ◽

...

Keyword(s):

Neural Network ◽

Machine Learning ◽

Convolutional Neural Network ◽

Prediction Accuracy ◽

Demand Forecasting ◽

Forecasting Model ◽

Learning Models ◽

Bike Sharing ◽

Demand Forecasting Model ◽

Machine Learning Models

Bike-sharing systems are becoming popular and generate a large volume of trajectory data. In a bike-sharing system, users can borrow and return bikes at different stations. In particular, a bike-sharing system will be affected by weather, the time period, and other dynamic factors, which challenges the scheduling of shared bikes. In this article, a new shared-bike demand forecasting model based on dynamic convolutional neural networks, called SDF , is proposed to predict the demand of shared bikes. SDF chooses the most relevant weather features from real weather data by using the Pearson correlation coefficient and transforms them into a two-dimensional dynamic feature matrix, taking into account the states of stations from historical data. The feature information in the matrix is extracted, learned, and trained with a newly proposed dynamic convolutional neural network to predict the demand of shared bikes in a dynamical and intelligent fashion. The phase of parameter update is optimized from three aspects: the loss function, optimization algorithm, and learning rate. Then, an accurate shared-bike demand forecasting model is designed based on the basic idea of minimizing the loss value. By comparing with classical machine learning models, the weight sharing strategy employed by SDF reduces the complexity of the network. It allows a high prediction accuracy to be achieved within a relatively short period of time. Extensive experiments are conducted on real-world bike-sharing datasets to evaluate SDF. The results show that SDF significantly outperforms classical machine learning models in prediction accuracy and efficiency.

Download Full-text

Automatic electronic invoice classification using machine learning models

10.20944/preprints202010.0057.v1 ◽

2020 ◽

Author(s):

Chiara Bardelli ◽

Alessandro Rondinelli ◽

Ruggero Vecchio ◽

Silvia Figini

Keyword(s):

Machine Learning ◽

Prediction Accuracy ◽

Intelligent System ◽

Classification Problem ◽

The Hierarchical Structure ◽

Input Variables ◽

Economic Nature ◽

Financial Transactions ◽

A Company ◽

Machine Learning Models

Electronic invoicing has become mandatory for Italian companies since January 2019. Invoices are structured in a predefined xml template where the information reported can be easily extracted and analyzed. The main aim of this paper is to exploit the information structured in electronic invoices to build an intelligent system which can facilitate accountants work. More precisely, this contribution shows how it is possible to automate part of the accounting process: all sent or received invoices of a company are classified into specific codes which represent the economic nature of the the financial transactions. In order to classify data contained in the invoices a machine learning multiclass classification problem is proposed using as input variables the information of the invoices to predict two different target variables, account codes and the VAT codes, which composes a general ledger entry. Different approaches are compared in terms of prediction accuracy. The best performance is achieved considering the hierarchical structure of the account codes.

Download Full-text

Machine Learning Method for TOC Prediction: Taking Wufeng and Longmaxi Shales in the Sichuan Basin, Southwest China as an Example

Geofluids ◽

10.1155/2021/6794213 ◽

2021 ◽

Vol 2021 ◽

pp. 1-13

Author(s):

Jia Rong ◽

Zongyuan Zheng ◽

Xiaorong Luo ◽

Chao Li ◽

Yuping Li ◽

...

Keyword(s):

Machine Learning ◽

Random Forest ◽

Shale Gas ◽

Sichuan Basin ◽

Prediction Accuracy ◽

Random Search ◽

Organic Carbon Content ◽

Learning Models ◽

The Sichuan Basin ◽

Machine Learning Models

The total organic carbon content (TOC) is a core indicator for shale gas reservoir evaluations. Machine learning-based models can quickly and accurately predict TOC, which is of great significance for the production of shale gas. Based on conventional logs, the measured TOC values, and other data of 9 typical wells in the Jiaoshiba area of the Sichuan Basin, this paper performed a Bayesian linear regression and applied a random forest machine learning model to predict TOC values of the shale from the Wufeng Formation and the lower part of the Longmaxi Formation. The results showed that the TOC value prediction accuracy was improved by more than 50% by using the well-trained machine learning models compared with the traditional Δ Log R method in an overmature and tight shale. Using the halving random search cross-validation method to optimize hyperparameters can greatly improve the speed of building the model. Furthermore, excluding the factors that affect the log value other than the TOC and taking the corrected data as input data for training could improve the prediction accuracy of the random forest model by approximately 5%. Data can be easily updated with machine learning models, which is of primary importance for improving the efficiency of shale gas exploration and development.

Download Full-text