Build Machine Learning Models Research Articles

Predicting Binding Affinity Between MHC-I Receptor and Peptides Based on Molecular Docking and Protein-peptide Interaction Interface Characteristics

Background:Predicting protein-peptide binding affinity is one of the leading research subjects in peptide drug design and repositioning. In previous studies, models constructed by researchers just used features of peptide structures. These features had limited information and could not describe the proteinpeptide interaction mode. This made models and predicted results lack interpretability in pharmacy and biology, which led to the protein-peptide interaction mode not being reflected. Therefore, it was of little significance for the design of peptide drugs.Objective:Considering the protein-peptide interaction mode, we extracted protein-peptide interaction interface characteristics and built machine learning models to improve the performance and enhance the interpretability of models.Methods:Taking MHC-I protein and its binding peptides as the research object, protein-peptide complexes were obtained by molecular docking, and 94 protein-peptide interaction interface characteristics were calculated. Then ten important features were selected using recursive feature elimination to construct SVR, RF, and MLP models to predict protein-peptide binding affinity.Results:The MAE of the SVR, RF and MLP models constructed using protein-peptide interaction interface characteristics are 0.2279, 0.2939 and 0.2041, their MSE are 0.1289, 0.1308 and 0.0780, and their R2 reached 0.8711, 0.8692 and 0.9220, respectively.Conclusion:The model constructed using protein-peptide interaction interface characteristics showed better prediction results. The key features for predicting protein-peptide binding affinity are the bSASA of negatively charged species, hydrogen bond acceptor, hydrophobic group, planarity, and aromatic ring.

Ensemble machine learning model trained on a new synthesized dataset generalizes well for stress prediction using wearable devices

Introduction:Advances in wearable sensor technology have enabled the collection of biomarkers that may correlate with levels of elevated stress. While significant research has been done in this domain, specifically in using machine learning to detect elevated levels of stress, the challenge of producing a machine learning model capable of generalizing well for use on new, unseen data remain. Acute stress response has both subjective, psychological and objectively measurable, biological components that can be expressed differently from person to person, further complicating the development of a generic stress measurement model. Another challenge is the lack of large, publicly available datasets labeled for stress response that can be used to develop robust machine learning models. In this paper, we first investigate the generalization ability of models built on datasets containing a small number of subjects, recorded in single study protocols. Next, we propose and evaluate methods combining these datasets into a single, large dataset to study the generalization capability of machine learning models built on larger datasets. Finally, we propose and evaluate the use of ensemble techniques by combining gradient boosting with an artificial neural network to measure predictive power on new, unseen data. In favor of reproducible research and to assist the community advance the field, we make all our experimental data and code publicly available through Github at https://github.com/xalentis/Stress. This paper’s in-depth study of machine learning model generalization for stress detection provides an important foundation for the further study of stress response measurement using sensor biomarkers, recorded with wearable technologies. Methods:Sensor biomarker data from six public datasets were utilized in this study. Exploratory data analysis was performed to understand the physiological variance between study subjects, and the complexity it introduces in building machine learning models capable of detecting elevated levels of stress on new, unseen data. To test model generalization, we developed a gradient boosting model trained on one dataset (SWELL), and tested its predictive power on two datasets previously used in other studies (WESAD, NEURO). Next, we merged four small datasets, i.e. (SWELL, NEURO, WESAD, UBFC-Phys), to provide a combined total of 99 subjects, and applied feature engineering to generate additional features utilizing statistical summaries, with sliding windows of 25 s. We name this large dataset, StressData. In addition, we utilized random sampling on StressData combined with another dataset (EXAM) to build a larger training dataset consisting of 200 synthesized subjects, which we name SynthesizedStressData. Finally, we developed an ensemble model that combines our gradient boosting model with an artificial neural network, and tested it using Leave-One-Subject-Out (LOSO) validation, and on two additional, unseen publicly available stress biomarker datasets (WESAD and Toadstool). Results:Our results show that previous models built on datasets containing a small number (<50) of subjects, recorded in single study protocols, cannot generalize well to new, unseen datasets. Our presented methodology for generating a large, synthesized training dataset by utilizing random sampling to construct scenarios closely aligned with experimental conditions demonstrate significant benefits. When combined with feature-engineering and ensemble learning, our method delivers a robust stress measurement system capable of achieving 85% predictive accuracy on new, unseen validation data, achieving a 25% performance improvement over single models trained on small datasets. The resulting model can be used as both a classification or regression predictor for estimating the level of perceived stress, when applied on specific sensor biomarkers recorded using a wearable device, while further allowing researchers to construct large, varied datasets for training machine learning models that closely emulate their exact experimental conditions. Conclusion:Models trained on small, single study protocol datasets do not generalize well for use on new, unseen data and lack statistical power. Machine learning models trained on a dataset containing a larger number of varied study subjects capture physiological variance better, resulting in more robust stress detection. Feature-engineering assists in capturing these physiological variance, and this is further improved by utilizing ensemble techniques by combining the predictive power of different machine learning models, each capable of learning unique signals contained within the data. While there is a general lack of large, labeled public datasets that can be utilized for training machine learning models capable of accurately measuring levels of acute stress, random sampling techniques can successfully be applied to construct larger, varied datasets from these smaller sample datasets, for building robust machine learning models.

Comprehensive Decision-Making Guide Predicting Colleges Based on User Profile Using Ensemble Machine Learning Model

Abstract: Making the right career decision is very importance for everyone. This paper aims to build a model that can help students to pick right colleges based on their profiles. For approximate prediction we plane build machine learning model in order to provide expected results. The datasets contain details about student (like Entrance Exam Score, 10th score, 12th score, cast etc.) and college (college Name, previous two-year cut-off etc.), for this various algorithm have been used i.e., Ensemble Machine learning, Linear Regression, Neural Network, Random Forest Decision Tree and Data Mining Technique for gathering the relevance data. This will assist student to know in advance if they get a chance to get recognized.

Addressing the data bottleneck in medical deep learning models using a human-in-the-loop machine learning approach

Any machine learning (ML) model is highly dependent on the data it uses for learning, and this is even more important in the case of deep learning models. The problem is a data bottleneck, i.e. the difficulty in obtaining an adequate number of cases and quality data. Another issue is improving the learning process, which can be done by actively introducing experts into the learning loop, in what is known as human-in-the-loop (HITL) ML. We describe an ML model based on a neural network in which HITL techniques were used to resolve the data bottleneck problem for the treatment of pancreatic cancer. We first augmented the dataset using synthetic cases created by a generative adversarial network. We then launched an active learning (AL) process involving human experts as oracles to label both new cases and cases by the network found to be suspect. This AL process was carried out simultaneously with an interactive ML process in which feedback was obtained from humans in order to develop better synthetic cases for each iteration of training. We discuss the challenges involved in including humans in the learning process, especially in relation to human–computer interaction, which is acquiring great importance in building ML models and can condition the success of a HITL approach. This paper also discusses the methodological approach adopted to address these challenges.

NIMG-39. BENCHMARKING MULTI-LABEL PREDICTION OF TARGET GENE MUTATION STATUS IN GLIOBLASTOMA USING RADIOGENOMICS AND MULTI-PARAMETRIC MR IMAGING

Abstract PURPOSE In vivo imaging-based prediction of gene mutation status can significantly contribute to the diagnosis and treatment of glioblastoma. Radiogenomics provides a non-invasive approach to capturing the phenotypical characteristics of molecular alterations. We build machine learning (ML) models on radiogenomics and MR imaging data for benchmarking the prediction of multiple target genes in a single institution dataset. METHODS Multi-parametric MRI scans (T1, T1-Gd, T2, T2-FLAIR, DSC, DTI) of 286 individuals with glioblastoma were used to extract radiomics features, including histograms, morphological and textural descriptors. Genetic markers were obtained through a targeted next generation sequencing panel. We built ML models to predict mutation status in two settings: (a) 4 frequently-occurring mutations (EGFR, PTEN, NF1, TP53), (b) 13 mutations including some that are infrequent (PDGFRA, MET, FGFR2, PIK3CA, PIK3R1, BRAF, MDM4, CDKN2A). The models include k-nearest neighbor classifiers, fully-connected neural networks w/o dimensionality reduction trained on radiomics data, convolutional (CNN) and self-attention-based (ViT) networks trained on MR imaging data. All models were trained with 5-fold cross validation on discovery cohort (n=257) and tested on replication cohort (n=29). RESULTS Multi-label prediction models achieved highest accuracies of 75.9% / 88.6% for 4 / 13 mutations using radiomics features. Models trained with radiomics features have a marginally better average F1 score than CNNs and ViTs trained on MR imaging data (by 4.1% / 4.6% for 4 / 13 mutations). CONCLUSION We had modest success in finding highly distinct MR signatures of multiple target mutations in glioblastoma using multi-label ML methods, presumably because of the small sample size and the molecular, the spatial heterogeneity of the tumor, and the difficulty of simultaneously learning multiple imaging signatures. This highlights the importance of pooling and harmonizing radiogenomics data from multiple institutions to increase the sample size, and of incorporating biological domain knowledge into ML models.

High-resolution magnetic resonance imaging-based radiomic features aid in selecting endovascular candidates among patients with cerebral venous sinus thrombosis

ObjectivesCerebral venous sinus thrombosis (CVST) can cause sinus obstruction and stenosis, with potentially fatal consequences. High-resolution magnetic resonance imaging (HRMRI) can diagnose CVST qualitatively, although quantitative screening methods are lacking for patients refractory to anticoagulation therapy and who may benefit from endovascular treatment (EVT). Thus, in this study, we used radiomic features (RFs) extracted from HRMRI to build machine learning models to predict response to drug therapy and determine the appropriateness of EVT.Materials and methodsRFs were extracted from three-dimensional T1-weighted motion-sensitized driven equilibrium (MSDE), T2-weighted MSDE, T1-contrast, and T1-contrast MSDE sequences to build radiomic signatures and support vector machine (SVM) models for predicting the efficacy of standard drug therapy and the necessity of EVT.ResultsWe retrospectively included 53 patients with CVST in a prospective cohort study, among whom 14 underwent EVT after standard drug therapy failed. Thirteen RFs were selected to construct the RF signature and CVST-SVM models. In the validation dataset, the sensitivity, specificity, and area under the curve performance for the RF signature model were 0.833, 0.937, and 0.977, respectively. The radiomic score was correlated with days from symptom onset, history of dyslipidemia, smoking, fibrin degradation product, and D-dimer levels. The sensitivity, specificity, and area under the curve for the CVST-SVM model in the validation set were 0.917, 0.969, and 0.992, respectively.ConclusionsThe CVST-SVM model trained with RFs extracted from HRMRI outperformed the RF signature model and could aid physicians in predicting patient responses to drug treatment and identifying those who may require EVT.

A knowledge graph-based data harmonization framework for secondary data reuse

Background and objectiveThe adoption of new technologies in clinical care systems has propitiated the availability of a great amount of valuable data. However, this data is usually heterogeneous, requiring its harmonization to be integrated and analysed. We propose a semantic-driven harmonization framework that (1) enables the meaningful sharing and integration of healthcare data across institutions and (2) facilitates the analysis and exploitation of the shared data. MethodsThe framework includes an ontology-based common data model (i.e. SCDM), a data transformation pipeline and a semantic query system. Heterogeneous datasets, mapped to different terminologies, are integrated by using an ontology-based infrastructure rooted in a top-level ontology. A graph database is generated by using these mappings, and web-based semantic query system facilitates data exploration. ResultsSeveral datasets from different European institutions have been integrated by using the framework in the context of the European H2020 Precise4Q project. Through the query system, data scientists were able to explore data and use it for building machine learning models. ConclusionsThe flexible data representation using RDF, together with the formal semantic underpinning provided by the SCDM, have enabled the semantic integration, query and advanced exploitation of heterogeneous data in the context of the Precise4Q project.

Revealing the impact of TOX3 on osteoarthritis: insights from bioinformatics.

Osteoarthritis, a prevalent long-term condition of the joints, primarily impacts older individuals, resulting in discomfort, restrictions in mobility, and a decrease in overall well-being. Although Osteoarthritis is widely spread, there is a lack of successful interventions to stop the advancement of the condition. Numerous signaling pathways have been emphasized in recent research on Osteoarthritis, yet the diagnostic significance of numerous genes has not been investigated. To identify genes that were expressed differently in osteoarthritis, we utilized the Gene Expression Omnibus database. To identify marker genes, we built machine learning models including Least Absolute Shrinkage and Selection Operator and Random Forest. We categorized Osteoarthritis samples and performed immune cell infiltration analysis based on the expression patterns of these characteristic genes. Both the Least Absolute Shrinkage and Selection Operator and Random Forest models selected six marker genes (TOX3, ARG1, CST7, RERGL, COL11A1, NCRNA00185) out of a total of 17 differentially expressed genes. The osteoarthritis samples were categorized into two groups, namely a high expression group and a low expression group, based on the median levels of TOX3 expression. Comparative analysis of these groups identified 85 differentially expressed genes, showing notable enrichment in pathways related to lipid metabolism in the group with high expression. Analysis of immune cell infiltration revealed noticeable differences in immune profiles among the two groups. The group with high expression of TOX3 showed a notable increase in Mast cells and Type II IFN Response, whereas B cells, Cytolytic activity, Inflammation-promoting cells, NK cells, pDCs, T cell co-inhibition, Th1 cells, and Th2 cells were significantly decreased. We constructed a ceRNA network for TOX3, revealing 57 lncRNAs and 18 miRNAs involved in 57 lncRNA-miRNA interactions, and 18 miRNA-mRNA interactions with TOX3. Validation of TOX3 expression was confirmed using an external dataset (GSE29746), revealing a notable increase in Osteoarthritis samples. In conclusion, our study presents a comprehensive analysis identifying TOX3 as a potential feature gene in Osteoarthritis. The distinct immune profiles and involvement in fat metabolism pathways associated with TOX3 expression suggest its significance in Osteoarthritis pathogenesis. The study establishes a basis for comprehending the intricate correlation between characteristic genes and Osteoarthritis, as well as for the formulation of individualized therapeutic approaches.

A Data-centric AI Framework for Automating Exploratory Data Analysis and Data Quality Tasks

Democratisation of machine learning (ML) has been an important theme in the research community for the last several years with notable progress made by the model-building community with automated machine learning models. However, data play a central role in building ML models and there is a need to focus on data-centric AI innovations. In this article, we first map the steps taken by data scientists for the data preparation phase and identify open areas and pain points via user interviews. We then propose a framework and four novel algorithms for exploratory data analysis and data quality for AI steps addressing the pain points from user interviews. We also validate our algorithms with open source datasets and show the effectiveness of our proposed methods. Next, we build a tool that automatically generates python code encompassing the above algorithms and study the usefulness of these algorithms via two user studies with data scientists. We observe from the first study results that the participants who used the tool were able to gain 2× productivity and 6% model improvement over the control group. The second study is performed in a more realistic environment to understand how the tool would be used in real-world scenarios. The results from this study are coherent with the first study and show an average of 30–50% of time savings that can be attributed to the tool.

Depression and anxiety have distinct and overlapping language patterns: Results from a clinical interview.

Depression has been associated with heightened first-person singular pronoun use (I-usage; e.g., "I," "my") and negative emotion words. However, past research has relied on nonclinical samples and nonspecific depression measures, raising the question of whether these features are unique to depression vis-à-vis frequently co-occurring conditions, especially anxiety. Using structured questions about recent life changes or difficulties, we interviewed a sample of individuals with varying levels of depression and anxiety (N = 486), including individuals in a major depressive episode (n = 228) and/or diagnosed with generalized anxiety disorder (n = 273). Interviews were transcribed to provide a natural language sample. Analyses isolated language features associated with gold standard, clinician-rated measures of depression and anxiety. Many language features associated with depression were in fact shared between depression and anxiety. Language markers with relative specificity to depression included I-usage, sadness, and decreased positive emotion, while negations (e.g., "not," "no"), negative emotion, and several emotional language markers (e.g., anxiety, stress, depression) were relatively specific to anxiety. Several of these results were replicated using a self-report measure designed to disentangle components of depression and anxiety. We next built machine learning models to detect severity of common and specific depression and anxiety using only interview language. Individuals' speech characteristics during this brief interview predicted their depression and anxiety severity, beyond other clinical and demographic variables. Depression and anxiety have partially distinct patterns of expression in spoken language. Monitoring of depression and anxiety severity via language can augment traditional assessment modalities and aid in early detection. (PsycInfo Database Record (c) 2023 APA, all rights reserved).

A machine learning approach to graduate admissions and the role of letters of recommendation.

The graduate admissions process is time-consuming, subjective, and complicated by the need to combine information from diverse data sources. Letters of recommendation (LORs) are particularly difficult to evaluate and it is unclear how much impact they have on admissions decisions. This study addresses these concerns by building machine learning models to predict admissions decisions for two STEM graduate programs, with a focus on examining the contribution of LORs in the decision-making process. We train our predictive models leveraging information extracted from structured application forms (e.g., undergraduate GPA, standardized test scores, etc.), applicants' resumes, and LORs. A particular challenge in our study is the different modalities of application data (i.e., text vs. structured forms). To address this issue, we converted the textual LORs into features using a commercial natural language processing product and a manual rating process that we developed. By analyzing the predictive performance of the models using different subsets of features, we show that LORs alone provide only modest, but useful, predictive signals to admission decisions; the best model for predicting admissions decisions utilized both LOR and non-LOR data and achieved 89% accuracy. Our experiments demonstrate promising results in the utility of automated systems for assisting with graduate admission decisions. The findings confirm the value of LORs and the effectiveness of our feature engineering methods from LOR text. This study also assesses the significance of individual features using the SHAP method, thereby providing insight into key factors affecting graduate admission decisions.

Pay Later Risk Management: A Review of FMECA and Potential Customer Prediction Frameworks Through the Application of Machine Learning

The development of technology continues to develop and gradually change the way people buy such as on online shopping sites. The increase in internet use, especially in the use of E Commerce, has given birth to great potential in the market, especially in Indonesia. These changes prompted the birth of various payment methods. One of them is Pay Later. 27% of the 3560 samples decided to use Pay Later with all the conveniences offered. However, the development of Pay Later is not synchronized with good risk management. The use of Pay Later, which is not targeted at the right consumers, causes PT. XYZ suffered losses due to 22.37% of users defaulting on Pay Later installments. The purpose of this study is to reduce Pay Later default users by answering what factors cause consumers to default. To support this study, the authors used FMECA, Cause Effect Diagrams and conducted tests using Machine Learning to improve company efficiency. Through critical matrix analysis, the author gets 3 priority failure modes, Users default, users disappear, and users experience payment delays. In solving the problems in this study, the authors provide recommendations in the form of a new framework in the form of analyzing the best Pay Later offers by analyzing consumer behavior patterns in an E Commerce by utilizing Machine Learning. However, future research will need to be conducted correlation analysis and static testing in testing attribute correlation before testing algorithms when building machine learning models. The authors also suggested comparing using other methods to improve risk management in this study.

Predicting rock hardness using Gaussian weighted moving average filter on borehole data and machine learning

A comprehensive understanding of the hardness of ore being handled and processed in a mining operation can significantly improve operational efficiencies. This is feasible by providing valuable data to support decision-making through the mining value chain (drilling, blasting, loading, comminution). This study presents the results of a machine learning (ML) approach for rock hardness prediction using rock’s geophysical and geochemical features. Core samples from several mine sites were logged using a multi-sensor core logging (MSCL) system. Measurements include ultrasonic P- and S-wave velocity, elemental concentration via portable X-Ray fluorescence analyzers (pXRF), and Leeb rebound hardness, measured every 30 cm along 564 m of core samples. K-Means and PCA were used for better interpretation of the data. Supervised ML models (XGBoost and Random Forest) were utilized to predict rock hardness using the elemental concentrations and ultrasonic velocities as predictors. Since the data was collected automatically with predefined intervals, some of the measurement points were near fractures or veins. The Gaussian weighted moving average (WMA) was used to smooth out variations in geochemistry or hardness caused by local features that do not reflect the overall rock characteristics. This approach is effective for building ML models to become less susceptible to local rock features. It was concluded that the rock hardness could be effectively predicted using only geochemistry, and the process of collecting P- and S-wave velocity for hardness prediction can be skipped.

Understanding Partnership Formation and Repeated Contributions in Federated Learning: An Analytical Investigation

Limited access to large-scale data is a key obstacle to building machine learning (ML) applications in practice, partly due to a reluctance of information exchange among data owners out of privacy and data security concerns. To address this “information silo” problem, federated learning (FL) techniques have been proposed to enable decentralized model training via an orchestrating central server and have received increasing attention in several industries (including healthcare and finance). Despite its superior privacy protection property, adoption of FL is limited by a lack of systematic understanding of its underlying economics. In this paper, we take an analytical approach to answer two questions: (1) when do data owners prefer to form a FL partnership over building ML models by themselves and (2) how can different contractual mechanisms be used to promote repeated contributions to FL (the cooperative outcome that benefits all participants). We formulate an iterated prisoner’s dilemma (IPD) model that accounts for unique FL characteristics, including the specification of the payoff matrix and the involvement of a central server to sanction noncooperation. We find that partnership formation requires participants to be not too forward-looking in temporal preferences, which is contrary to the conventional wisdom in IPD. Furthermore, to promote repeated contributions, it is insufficient to only rely on penalties imposed by the central server or by participants for noncooperation, but a combination of both is enough. Our work advances theoretical understanding of the economics of FL and provides prescriptive insights that can inform FL participant selection and contract design. This paper was accepted by D. J. Wu, information systems. Funding: This work was partially supported by Cisco Research. Supplemental Material: The online appendices are available at https://doi.org/10.1287/mnsc.2023.00611 .

Prediction of Surgical Upstaging Risk of Ductal Carcinoma In Situ Using Machine Learning Models.

The purpose of this study was to build machine learning models to predict surgical upstaging risk of ductal carcinoma in situ (DCIS) to invasive cancer and to compare model performance to eligibility criteria used by the Comparison of Operative versus Monitoring and Endocrine Therapy (COMET) active surveillance trial. Medical records were retrospectively reviewed of all women with DCIS at core-needle biopsy who underwent surgery from 2007 to 2016 at an academic medical center. Multivariable regression and machine learning models were developed to evaluate upstaging-related features and their performance was compared with that achieved using the COMET trial eligibility criteria. Of 1387 women (mean age, 57 years; range, 27-89 years), the upstaging rate of DCIS was 17% (235/1387). On multivariable analysis, upstaging-associated features were presentation of DCIS as a palpable area of concern, imaging finding of a mass, and nuclear grades 2 or 3 at biopsy (P < 0.05). If COMET trial eligibility criteria were applied to our study cohort, then 496 women (42%, 496/1175) would have been eligible for the trial, with an upstaging rate of 12% (61/496). Of the machine learning models, none had a significantly lower upstaging rate than 12%. However, if using the models to determine eligibility, then a significantly larger proportion of women (56%-87%) would have been eligible for active surveillance. Use of machine learning models to determine eligibility for the COMET trial identified a larger proportion of women eligible for surveillance compared with current eligibility criteria while maintaining similar upstaging rates.

Predicting open education competency level: A machine learning approach

This article aims to study open education competency data through machine learning models to determine whether models can be built on decision rules using the features from the students' perceptions and classify them by the level of competency. Data was collected from a convenience sample of 326 students from 26 countries using the eOpen instrument. Based on a quantitative research approach, we analyzed the eOpen data using two machine learning models considering these findings: 1) derivation of decision rules from students' perceptions of knowledge, skills, and attitudes or values related to open education to predict their competence level using Decision Trees and Random Forests models, 2) analysis of the prediction errors in the machine learning models to find bias, and 3) description of decision trees from the machine learning models to understand the choices that both models made to predict the competency levels. The results confirmed our hypothesis that the students' perceptions of their knowledge, skills, and attitudes or values related to open education and its sub-competencies produced satisfactory data for building machine learning models to predict the participants' competency levels.

SportsTables: A New Corpus for Semantic Type Detection (Extended Version)

Table corpora such as VizNet or TURL which contain annotated semantic types per column are important to build machine learning models for the task of automatic semantic type detection. However, there is a huge discrepancy between corpora and real-world data lakes since they contain a huge fraction of numerical data which are not present in existing corpora. Hence, in this paper, we introduce a new corpus that contains a much higher proportion of numerical columns than existing corpora. To reflect the distribution in real-world data lakes, our corpus SportsTables has on average approx. 86% numerical columns, posing new challenges to existing semantic type detection models which have mainly targeted non-numerical columns so far. To demonstrate this effect, we show in this extended version paper of [18] the results of an extensive study using four different state-of-the-art approaches for semantic type detection on our new corpus. Overall, the results demonstrate significant performance differences in predicting semantic types for textual and numerical data.

A Comparative Investigation of Random Forest Regression and Artificial Neural Networks for Predicting Crack Growth Life of a Fighter Aircraft Wing Joint Under Spectrum Loading

Estimating fatigue life is challenging due to the input parameters’ statistical natures, such as the manufacturing process, scatter of service loads, microstructure, etc. Regarding fatigue life calculation in the aerospace industry, the importance of an accurate estimation becomes more critical due to strict safety, certification, service costs, and competitiveness regulations. The ability of soft computing methods to reveal complex relationships between multiple parameters and their computational speed could help predict fatigue life, especially in the service. This study compares random forest regression and artificial neural network methods to estimate the crack growth life of a fighter jet aircraft wing joint in terms of their computational time and accuracy. In addition, permutation feature importance and hyper parameter optimisation studies are conducted to extract essential features, investigate their effects on estimation performance, and fine-tune model parameters. The analysed joint is made of 7050-T7451 aluminium, widely used as a structural element in the aerospace industry. Since a hole is one of the major sources of stress concentration, and there may be many holes involved in any engineering structure, it is reasonable to assume that fatigue cracks may initiate at some of these holes during the service life of engineering structures. The crack type considered is a thru-crack around a hole, which is more severe than a corner crack. Load spectra are derived using the Fighter Aircraft Loading Standard for Fatigue (FALSTAFF) to calculate crack growth life. Considering particular service load conditions, ninety different spectra are developed, and the crack growth life of the joint is calculated based on linear elastic fracture mechanics correspondingly. Also, to simulate the worst-case scenario, friction between members and the retardation effect of load spectra are not considered when calculating crack growth life. Python’s Tensor Flow and Scikit-learn libraries are utilised to build machine learning models......

COVID-Net Biochem: an explainability-driven framework to building machine learning models for predicting survival and kidney injury of COVID-19 patients from clinical and biochemistry data

Since the World Health Organization declared COVID-19 a pandemic in 2020, the global community has faced ongoing challenges in controlling and mitigating the transmission of the SARS-CoV-2 virus, as well as its evolving subvariants and recombinants. A significant challenge during the pandemic has not only been the accurate detection of positive cases but also the efficient prediction of risks associated with complications and patient survival probabilities. These tasks entail considerable clinical resource allocation and attention. In this study, we introduce COVID-Net Biochem, a versatile and explainable framework for constructing machine learning models. We apply this framework to predict COVID-19 patient survival and the likelihood of developing Acute Kidney Injury during hospitalization, utilizing clinical and biochemical data in a transparent, systematic approach. The proposed approach advances machine learning model design by seamlessly integrating domain expertise with explainability tools, enabling model decisions to be based on key biomarkers. This fosters a more transparent and interpretable decision-making process made by machines specifically for medical applications. More specifically, the framework comprises two phases: In the first phase, referred to as the “clinician-guided design” phase, the dataset is preprocessed using explainable AI and domain expert input. To better demonstrate this phase, we prepared a benchmark dataset of carefully curated clinical and biochemical markers based on clinician assessments for survival and kidney injury prediction in COVID-19 patients. This dataset was selected from a patient cohort of 1366 individuals at Stony Brook University. Moreover, we designed and trained a diverse collection of machine learning models, encompassing gradient-based boosting tree architectures and deep transformer architectures, specifically for survival and kidney injury prediction based on the selected markers. In the second phase, called the “explainability-driven design refinement” phase, the proposed framework employs explainability methods to not only gain a deeper understanding of each model’s decision-making process but also to identify the overall impact of individual clinical and biochemical markers for bias identification. In this context, we used the models constructed in the previous phase for the prediction task and analyzed the explainability outcomes alongside a clinician with over 8 years of experience to gain a deeper understanding of the clinical validity of the decisions made. The explainability-driven insights obtained, in conjunction with the associated clinical feedback, are then utilized to guide and refine the training policies and architectural design iteratively. This process aims to enhance not only the prediction performance but also the clinical validity and trustworthiness of the final machine learning models. Employing the proposed explainability-driven framework, we attained 93.55% accuracy in survival prediction and 88.05% accuracy in predicting kidney injury complications. The models have been made available through an open-source platform. Although not a production-ready solution, this study aims to serve as a catalyst for clinical scientists, machine learning researchers, and citizen scientists to develop innovative and trustworthy clinical decision support solutions, ultimately assisting clinicians worldwide in managing pandemic outcomes.

Local Industrialization Based Lucrative Farming Using Machine Learning Technique

In recent times, agriculture have gained lot of attention of researchers. More precisely, crop prediction is trending topic for research as it leads agri-business to success or failure. Crop prediction totally rest on climatic and chemical changes. In the past which crop to promote was elected by rancher. All the decisions related to its cultivation, fertilizing, harvesting and farm maintenance was taken by rancher himself with his experience. But as we can see because of constant fluctuations in atmospheric conditions coming to any conclusion have become very tough. Picking correct crop to grow at right times under right circumstances can help rancher to make more business. To achieve what we cannot do manually we have started building machine learning models for it nowadays. To predict the crop deciding which parameters to consider and whose impact will be more on final decision is also equally important. For this we use feature selection models. This will alter the underdone data into more precise one. Though there have been various techniques to resolve this problem better performance is still desirable. In this research we have provided more precise &amp; optimum solution for crop prediction keeping Satara, Sangli, Kolhapur region of Maharashtra. Along with crop &amp; composts to increase harvest we are offering industrialization around so rancher can trade the yield &amp; earn more profit. The proposed solution is using machine learning algorithms like KNN, Random Forest, Naïve Bayes where Random Forest outperforms others so we are using it to build our final framework to predict crop.