Pre-trained Machine Learning Model Research Articles

Automated analysis of a novel object recognition test in mice using image processing and machine learning

The novel object recognition test (NORT) is one of the most commonly employed behavioral tests in experimental animals designed to evaluate an animal's interest in and recognition of novelty. However, manual procedures, which rely on researchers’ observations, prevent high throughput analysis. In this study, we developed an automated analysis method for NORT utilizing machine learning-assisted exploratory behavior detection. We recorded the exploratory behavior of the mice using a video camera. The coordinates of the mouse nose and tail base in recorded video files were detected using a pre-trained machine learning model, DeepLabCut. Each video was then segmented into frame images, which were categorized into "exploratory,” or "non-exploratory" frames based on manual observation. Mouse feature vectors were calculated as vectors from the nose to the vertices of the object and were utilized for SVM training. The trained SVM effectively detected exploratory behaviors, showing a strong correlation with human observer assessments. Upon application to NORT, the duration of mouse exploratory behavior towards objects predicted by the SVM exhibited a significant correlation with the assessments made by human observers. The novelty discrimination index derived from the SVM predictions also aligned well with that from human observations.

Use of a structure-borne sound-based in-process sensor system to identify Weld seam irregularities during electron beam welding

In this publication, an in-process quality assurance method for electron beam welding based on a structure-borne sound emission test for the detection of weld irregularities arising in the process is presented. For this purpose, different sheet materials, i.e., AISI 304, AZ31 and AlMg3, were welded in a butt-joint and the resulting process noises were recorded by means of two acoustic emission sensors specifically designed for structure-borne sound. During the welding experiments, typical irregularities, e.g. incidence points, pore lines and cracks, were deliberately induced. Subsequently, the recorded acoustic signals were examined with regard to defect-specific abnormalities. Various methods in the time and frequency domain as well as pre-trained machine learning models were used to analyze the acoustic emission data. The results show that the investigated irregularities can be identified and distinguished from other process emissions, eventually enabling a robust means of identification for weld seam irregularities and, thus, opening pathways towards cost-effective in-process quality control.

MPRAVarDB: an online database and web server for exploring regulatory effects of genetic variants.

Massively parallel reporter assay (MPRA) is an important technology for evaluating the impact of genetic variants on gene regulation. Here, we present MPRAVarDB, an online database and web server for exploring regulatory effects of genetic variants. MPRAVarDB harbors 18 MPRA experiments designed to assess the regulatory effects of genetic variants associated with GWAS loci, eQTLs, and genomic features, totaling 242818 variants tested more than 30 cell lines and 30 human diseases or traits. MPRAVarDB enables users to query MPRA variants by genomic region, disease and cell line, or any combination of these parameters. Notably, MPRAVarDB offers a suite of pretrained machine-learning models tailored to the specific disease and cell line, facilitating the prediction of regulatory variants. The user-friendly interface allows users to receive query and prediction results with just a few clicks. https://mpravardb.rc.ufl.edu.

TigeR: Tumor immunotherapy gene expression data analysis R package

AbstractImmunotherapy shows great promise for treating advanced cancers, but its effectiveness varies widely among different patients and cancer types. Identifying biomarkers and developing robust predictive models to discern which patients are most likely to benefit from immunotherapy is of great importance. In this context, we have developed the tumor immunotherapy gene expression R package (tigeR 1.0) to address the increasing need for effective tools to explore biomarkers and construct predictive models. tigeR encompasses four distinct yet closely interconnected modules. The Biomarker Evaluation module enables researchers to evaluate whether the biomarkers of interest are associated with immunotherapy response via built‐in or custom immunotherapy gene expression data. The Tumor Microenvironment Deconvolution module integrates 10 open‐source algorithms to obtain the proportions of different cell types within the tumor microenvironment, facilitating the investigation of the association between immune cell populations and immunotherapy response. The Prediction Model Construction module equips users with the ability to construct sophisticated prediction models using a range of built‐in machine‐learning algorithms. The Response Prediction module predicts the immunotherapy response for the patients from gene expression data using our pretrained machine learning models or public gene expression signatures. By providing these diverse functionalities, tigeR aims to simplify the process of analyzing immunotherapy gene expression data, thus making it accessible to researchers without advanced programming skills. The source code and example for the tigeR project can be accessed at http://github.com/YuLab-SMU/tigeR.

Ozone exceedance forecasting with enhanced extreme instance augmentation: A case study in Germany

Accurately forecasting ozone levels that exceed specific thresholds is pivotal for mitigating adverse effects on both the environment and public health. However, predicting such ozone exceedances remains challenging due to the infrequent occurrence of high-concentration ozone data. This research, leveraging data from 57 German monitoring stations from 1999 to 2018, introduces an Enhanced Extreme Instance Augmentation Random Forest (EEIA-RF) approach that significantly improves the prediction of days when the maximum daily 8-hour average ozone concentrations exceed 120μg/m3. A pre-trained machine learning model is used to generate additional high-concentration data, which, combined with selectively reduced low-concentration data, forms a new dataset for training a refined model. This method achieved an improvement of at least 8% in the accuracy of predicting days with ozone exceedances across Germany. Our experiment underscores the approach’s value in enhancing atmospheric modeling and supporting public health advisories and environmental policy-making related to ozone pollution.

Designing of potential polymer donors with lower reorganization energy and chemical space exploration in search of materials for efficient organic solar cells

Machine learning with fascinating material predicting capabilities has emerged as a growing throughput artificial intelligence technology to accelerate the design, screening and prediction of material properties. In this study, chemical descriptors are employed to train machine learning (ML) models for the designing of potential polymer donor materials aimed at enhancing the efficiency of organic solar cells (OSCs). To achieve this, ten ML models are evaluated. Among all, the random forest regressor model has shown better predictive ability. In addition, to quantify the importance of features, the numerical scores of the significance of the features are determined through Feature importance analysis. Moreover, 10k new polymers are generated employing the python-based method and their reorganization energies are further predicted using a pre-trained ML model. The assemblage of polymers is done on the basis of the reorganization energy, those with lower reorganization energy being preferred. Furthermore, the synthetic accessibility and thus the structural diversity among the chosen polymers donors are examined, revealing a structural difference among the chosen polymers. This work paves a way for the designing and screening of efficient semiconductors with low reorganization energy to increase the efficiency of OSCs.

Unraveling Reactivity Origin of Oxygen Reduction at High-Entropy Alloy Electrocatalysts with a Computational and Data-Driven Approach.

High-entropy alloys (HEAs), characterized as compositionally complex solid solutions with five or more metal elements, have emerged as a novel class of catalytic materials with unique attributes. Because of the remarkable diversity of multielement sites or site ensembles stabilized by configurational entropy, human exploration of the multidimensional design space of HEAs presents a formidable challenge, necessitating an efficient, computational and data-driven strategy over traditional trial-and-error experimentation or physics-based modeling. Leveraging deep learning interatomic potentials for large-scale molecular simulations and pretrained machine learning models of surface reactivity, our approach effectively rationalizes the enhanced activity of a previously synthesized PdCuPtNiCo HEA nanoparticle system for electrochemical oxygen reduction, as corroborated by experimental observations. We contend that this framework deepens our fundamental understanding of the surface reactivity of high-entropy materials and fosters the accelerated development and synthesis of monodisperse HEA nanoparticles as a versatile material platform for catalyzing sustainable chemical and energy transformations.

An innovative approach to design readily synthesizable polymers for all-polymer solar cells

The communalization of all-polymer solar cells depends on the cost of active layer materials. We have introduced a framework to find the easily synthesizable polymers. Breaking Retrosynthetically Interesting Chemical Substructures (BRICS) method is used to generate a large database of polymers and synthetic accessibility of generated polymers is predicted. The generated database is visualized using various methods. Energy levels of polymers are also predicted using pretrained machine learning models. Polymers are screened on the basis of predicted properties. Library of polymers are displayed using the T-distributed Stochastic Neighbor Embedding (t-SNE) visualization. Structure Activity Landscape Index (SALI) visualization is also used. A significant change is observed in synthetic accessibility score on structural changes. The histagradient boosting regressor is used to predict the energy levels of polymers that energy levels play significant role in the selection of materials for organic photovoltaic cells. Synthetic accessibility of polymers is analyzed and a significant number of polymers are easy to synthesize. Thirty polymers are selected through screening process that are potential candidates for organic photovoltaic cells.

ZTCloudGuard: Zero Trust Context-Aware Access Management Framework to Avoid Medical Errors in the Era of Generative AI and Cloud-Based Health Information Ecosystems

Managing access between large numbers of distributed medical devices has become a crucial aspect of modern healthcare systems, enabling the establishment of smart hospitals and telehealth infrastructure. However, as telehealth technology continues to evolve and Internet of Things (IoT) devices become more widely used, they are also increasingly exposed to various types of vulnerabilities and medical errors. In healthcare information systems, about 90% of vulnerabilities emerge from medical error and human error. As a result, there is a need for additional research and development of security tools to prevent such attacks. This article proposes a zero-trust-based context-aware framework for managing access to the main components of the cloud ecosystem, including users, devices, and output data. The main goal and benefit of the proposed framework is to build a scoring system to prevent or alleviate medical errors while using distributed medical devices in cloud-based healthcare information systems. The framework has two main scoring criteria to maintain the chain of trust. First, it proposes a critical trust score based on cloud-native microservices for authentication, encryption, logging, and authorizations. Second, a bond trust scoring system is created to assess the real-time semantic and syntactic analysis of attributes stored in a healthcare information system. The analysis is based on a pre-trained machine learning model that generates the semantic and syntactic scores. The framework also takes into account regulatory compliance and user consent in the creation of the scoring system. The advantage of this method is that it applies to any language and adapts to all attributes, as it relies on a language model, not just a set of predefined and limited attributes. The results show a high F1 score of 93.5%, which proves that it is valid for detecting medical errors.

External Validation of the Machine Learning-Based Thermographic Indices for Rheumatoid Arthritis: A Prospective Longitudinal Study.

External validation is crucial in developing reliable machine learning models. This study aimed to validate three novel indices-Thermographic Joint Inflammation Score (ThermoJIS), Thermographic Disease Activity Index (ThermoDAI), and Thermographic Disease Activity Index-C-reactive protein (ThermoDAI-CRP)-based on hand thermography and machine learning to assess joint inflammation and disease activity in rheumatoid arthritis (RA) patients. A 12-week prospective observational study was conducted with 77 RA patients recruited from rheumatology departments of three hospitals. During routine care visits, indices were obtained at baseline and week 12 visits using a pre-trained machine learning model. The performance of these indices was assessed cross-sectionally and longitudinally using correlation coefficients, the area under the receiver operating curve (AUROC), sensitivity, specificity, and positive and negative predictive values. ThermoDAI and ThermoDAI-CRP correlated with CDAI, SDAI, and DAS28-CRP cross-sectionally (ρ = 0.81; ρ = 0.83; ρ = 0.78) and longitudinally (ρ = 0.55; ρ = 0.61; ρ = 0.60), all p < 0.001. ThermoDAI and ThermoDAI-CRP also outperformed Patient Global Assessment (PGA) and PGA + C-reactive protein (CRP) in detecting changes in 28-swollen joint counts (SJC28). ThermoJIS had an AUROC of 0.67 (95% CI, 0.58 to 0.76) for detecting patients with swollen joints and effectively identified patients transitioning from SJC28 > 1 at baseline visit to SJC28 ≤ 1 at week 12 visit. These results support the effectiveness of ThermoJIS in assessing joint inflammation, as well as ThermoDAI and ThermoDAI-CRP in evaluating disease activity in RA patients.

Unraveling dynamic protein structures by two-dimensional infrared spectra with a pretrained machine learning model

Dynamic protein structures are crucial for deciphering their diverse biological functions. Two-dimensional infrared (2DIR) spectroscopy stands as an ideal tool for tracing rapid conformational evolutions in proteins. However, linking spectral characteristics to dynamic structures poses a formidable challenge. Here, we present a pretrained machine learning model based on 2DIR spectra analysis. This model has learned signal features from approximately 204,300 spectra to establish a "spectrum-structure" correlation, thereby tracing the dynamic conformations of proteins. It excels in accurately predicting the dynamic content changes of various secondary structures and demonstrates universal transferability on real folding trajectories spanning timescales from microseconds to milliseconds. Beyond exceptional predictive performance, the model offers attention-based spectral explanations of dynamic conformational changes. Our 2DIR-based pretrained model is anticipated to provide unique insights into the dynamic structural information of proteins in their native environments.

Advancing Skin Cancer Prediction Using Ensemble Models

There are many different kinds of skin cancer, and an early and precise diagnosis is crucial because skin cancer is both frequent and deadly. The key to effective treatment is accurately classifying the various skin cancers, which have unique traits. Dermoscopy and other advanced imaging techniques have enhanced early detection by providing detailed images of lesions. However, accurately interpreting these images to distinguish between benign and malignant tumors remains a difficult task. Improved predictive modeling techniques are necessary due to the frequent occurrence of erroneous and inconsistent outcomes in the present diagnostic processes. Machine learning (ML) models have become essential in the field of dermatology for the automated identification and categorization of skin cancer lesions using image data. The aim of this work is to develop improved skin cancer predictions by using ensemble models, which combine numerous machine learning approaches to maximize their combined strengths and reduce their individual shortcomings. This paper proposes a fresh and special approach for ensemble model optimization for skin cancer classification: the Max Voting method. We trained and assessed five different ensemble models using the ISIC 2018 and HAM10000 datasets: AdaBoost, CatBoost, Random Forest, Gradient Boosting, and Extra Trees. Their combined predictions enhance the overall performance with the Max Voting method. Moreover, the ensemble models were fed with feature vectors that were optimally generated from the image data by a genetic algorithm (GA). We show that, with an accuracy of 95.80%, the Max Voting approach significantly improves the predictive performance when compared to the five ensemble models individually. Obtaining the best results for F1-measure, recall, and precision, the Max Voting method turned out to be the most dependable and robust. The novel aspect of this work is that skin cancer lesions are more robustly and reliably classified using the Max Voting technique. Several pre-trained machine learning models’ benefits are combined in this approach.

Object Detection for the Visually Impaired

This paper presents the design and development of a mobile application, built using Flutter, that leverages object detection to enhance the lives of visually impaired individuals. The application addresses a crucial challenge faced by this community – the lack of real-time information about their surroundings. We propose a solution that utilizes pre-trained machine learning models, potentially through TensorFlow Lite for on-device processing, to identify objects within the user's field of view captured by the smartphone camera. The application goes beyond simple object recognition. Detected objects are translated into natural language descriptions through text-to-speech functionality, providing crucial auditory cues about the environment. This real-time information stream empowers users to navigate their surroundings with greater confidence and independence. Accessibility is a core principle of this project. The user interface will be designed with compatibility for screen readers, ensuring seamless interaction for users who rely on assistive technologies. Haptic feedback mechanisms will be incorporated to provide non-visual cues and enhance the user experience. The ultimate goal of this project is to create a user-friendly and informative application that empowers visually impaired people to gain greater independence in their daily lives. The application has the potential to improve spatial awareness, foster a sense of security, and promote overall inclusion within society

Prediction of allowable compression load for notched composite laminates combining FEA simulation and machine learning

Determining the allowable compression load (ACL) of notched composite laminates (NCL) requires consideration of buckling, intra- and inter-laminar damage, which is a costly and repetitive task due to the high dimensional and nonlinear relation of ACL to the numerous design variables. In this work, a high-throughput finite element analysis (FEA) and machine learning (ML) combined approach is proposed to predict ACL of laminates. First, the high-throughput FEA model of NCL covering all of the design variables is generated, and the corresponding critical compression loads in terms of buckling, intra- and inter-laminar damage are obtained. Then, ML models are trained, with inputs being the design variables of NCL and outputs being the critical compression loads. ACL and initial failure mode are determined by the minimum compression load of the three failure modes. Moreover, transfer learning is introduced to laminates with new design variables of new ply number, notch shape and laminate type. By fine-tuning the pre-trained ML model using scarce new data, the fine-tuned models are suitable for new laminates. Consequently, ACL and initial failure mode of notched laminates with various design variables are predicted.

Building datasets to support information extraction and structure parsing from electronic theses and dissertations

Despite the millions of electronic theses and dissertations (ETDs) publicly available online, digital library services for ETDs have not evolved past simple search and browse at the metadata level. We need better digital library services that allow users to discover and explore the content buried in these long documents. Recent advances in machine learning have shown promising results for decomposing documents into their constituent parts, but these models and techniques require data for training and evaluation. In this article, we present high-quality datasets to train, evaluate, and compare machine learning methods in tasks that are specifically suited to identify and extract key elements of ETD documents. We explain how we construct the datasets by manual labeling the data or by deriving labeled data through synthetic processes. We demonstrate how our datasets can be used to develop downstream applications and to evaluate, retrain, or fine-tune pre-trained machine learning models. We describe our ongoing work to compile benchmark datasets and exploit machine learning techniques to build intelligent digital libraries for ETDs.

Efficient Automated Model for Predicting and Detecting the Heart Disease Through Machine Learning

Using machine learning, numerous researchers have created a variety of disease prediction models in the contemporary environment. By using prediction models early on, the death rate could be reduced. This work presents the design of an automated database diagnosis model using machine learning. Here, for study purposes, we chose one of the most common diseases, heart illness. This study presents the preparation of a database of patients with heart disease, followed by analysis in the produced database using a pre-trained machine learning model that was trained on the disease detection result that is displayed on the screen. To do the computation for prediction, a single logistic regression is utilized. Determining the risk of heart disease can be aided by early identification and prediction of the condition. Additionally, a comparison study might aid medical professionals in prescribing medication on time.

Probing the complexity of wood with computer vision: from pixels to properties.

We use data produced by industrial wood grading machines to train a machine learning model for predicting strength-related properties of wood lamellae from colour images of their surfaces. The focus was on samples of Norway spruce (Picea abies) wood, which display visible fibre pattern formations on their surfaces. We used a pre-trained machine learning model based on the residual network ResNet50 that we trained with over 15 000 high-definition images labelled with the indicating properties measured by the grading machine. With the help of augmentation techniques, we were able to achieve a coefficient of determination (R2) value of just over 0.9. Considering the ever-increasing demand for construction-grade wood, we argue that computer vision should be considered a viable option for the automatic sorting and grading of wood lamellae in the future.

Chemical library generation of polymer acceptors for organic solar cells with higher electron affinity

In this study, an intricate machine learning assisted framework is introduced for the designing of polymer acceptors. Machine learning (ML) models are trained to predict the electron affinity of polymers. Ten thousand new polymers are generated using the Breaking Retrosynthetically Interesting Chemical Substructures (BRICS) method and their electron affinity values are predicted using a pre-trained ML model. The selection of polymers is based on electron affinity, retaining those with higher electron affinity. The synthetic accessibility of the selected polymers is evaluated. Additionally, the structural similarity among the selected polymers is examined; results are indicating higher similarity between selected compounds. By efficiently identifying and optimizing new polymers, the approaches developed significantly enhance the likelihood of discovering superior materials for advanced applications.

Deep Learning Model for Prediction of Bronchopulmonary Dysplasia in Preterm Infants Using Chest Radiographs

Bronchopulmonary dysplasia (BPD) is common in preterm infants and may result in pulmonary vascular disease, compromising lung function. This study aimed to employ artificial intelligence (AI) techniques to help physicians accurately diagnose BPD in preterm infants in a timely and efficient manner. This retrospective study involves two datasets: a lung region segmentation dataset comprising 1491 chest radiographs of infants, and a BPD prediction dataset comprising 1021 chest radiographs of preterm infants. Transfer learning of a pre-trained machine learning model was employed for lung region segmentation and image fusion for BPD prediction to enhance the performance of the AI model. The lung segmentation model uses transfer learning to achieve a dice score of 0.960 for preterm infants with ≤\\documentclass[12pt]{minimal} \\usepackage{amsmath} \\usepackage{wasysym} \\usepackage{amsfonts} \\usepackage{amssymb} \\usepackage{amsbsy} \\usepackage{mathrsfs} \\usepackage{upgreek} \\setlength{\\oddsidemargin}{-69pt} \\begin{document}$$\\le$$\\end{document} 168 h postnatal age. The BPD prediction model exhibited superior diagnostic performance compared to that of experts and demonstrated consistent performance for chest radiographs obtained at ≤\\documentclass[12pt]{minimal} \\usepackage{amsmath} \\usepackage{wasysym} \\usepackage{amsfonts} \\usepackage{amssymb} \\usepackage{amsbsy} \\usepackage{mathrsfs} \\usepackage{upgreek} \\setlength{\\oddsidemargin}{-69pt} \\begin{document}$$\\le$$\\end{document} 24 h postnatal age, and those obtained at 25 to 168 h postnatal age. This study is the first to use deep learning on preterm chest radiographs for lung segmentation to develop a BPD prediction model with an early detection time of less than 24 h. Additionally, this study compared the model’s performance according to both NICHD and Jensen criteria for BPD. Results demonstrate that the AI model surpasses the diagnostic accuracy of experts in predicting lung development in preterm infants.

Ensuring cross-device portability of electromagnetic side-channel analysis for digital forensics

Investigation on smart devices has become an essential subdomain in digital forensics. The inherent diversity and complexity of smart devices pose a challenge to the extraction of evidence without physically tampering with it, which is often a strict requirement in law enforcement and legal proceedings. Recently, this has led to the application of non-intrusive Electromagnetic Side-Channel Analysis (EM-SCA) as an emerging approach to extract forensic insights from smart devices. EM-SCA for digital forensics is still in its infancy, and has only been tested on a small number of devices so far. Most importantly, the question still remains whether Machine Learning (ML) models in EM-SCA are portable across multiple devices to be useful in digital forensics, i.e., cross-device portability. This study experimentally explores this aspect of EM-SCA using a wide set of smart devices. The experiments using various iPhones and Nordic Semiconductor nRF52-DK devices indicate that the direct application of pre-trained ML models across multiple identical devices does not yield optimal outcomes (under 20 % accuracy in most cases). Subsequent experiments included collecting distinct samples of EM traces from all the devices to train new ML models with mixed device data; this also fell short of expectations (still below 20 % accuracy). This prompted the adoption of transfer learning techniques, which showed promise for cross-model implementations. In particular, for the iPhone 13 and nRF52-DK devices, applying transfer learning techniques resulted in achieving the highest accuracy, with accuracy scores of 98 % and 96 %, respectively. This result makes a significant advancement in the application of EM-SCA to digital forensics by enabling the use of pre-trained models across identical or similar devices.