Applying Machine Learning Tools Research Articles

Rates and Beaming Angles of Gamma-Ray Bursts Associated with Compact Binary Coalescences

Some, if not all, binary neutron star (BNS) coalescences, and a fraction of neutron star–black hole (NSBH) mergers, are thought to produce sufficient mass ejection to power gamma-ray bursts (GRBs). However, this fraction, as well as the distribution of beaming angles of BNS-associated GRBs, is poorly constrained from observation. Recent work applied machine learning tools to analyze GRB light curves observed by Fermi/Gamma-Ray Burst Monitor (GBM) and Swift/Burst Alert Telescope (BAT). GRBs were segregated into multiple distinct clusters, with the tantalizing possibility that one of them (BNS cluster) could be associated with BNSs and another (NSBH cluster) with NSBHs. As a proof of principle, assuming that all GRBs detected by Fermi/GBM and Swift/BAT associated with BNSs (NSBHs) lie in the BNS (NSBH) cluster, we estimate their rates (Gpc−3 yr−1). We compare these rates with corresponding BNS and NSBH rates estimated by the LIGO–Virgo–KAGRA (LVK) collaboration from the first three observing runs (O1, O2, O3). We find that the BNS rates are consistent with LVK’s rate estimates, assuming a uniform distribution of beaming fractions (f b ∈ [0.01, 0.1]). Conversely, using the LVK’s BNS rate estimates, assuming all BNS mergers produce GRBs, we are able to constrain the beaming angle distribution to θ j ∈ [0.°8, 33.°5] at 90% confidence. We similarly place limits on the fraction of GRB-bright NSBHs as f B ∈ [1.3%, 63%] (f B ∈ [0.4%, 15%]) with Fermi/GBM (Swift/BAT) data.

Moho depth model of South America from a machine learning approach

Crustal structure models play an important role in the characterization of seismogenic zones, in the regional delimitation of geological provinces and particularly, in understanding the genesis and evolution of sedimentary basins. Despite the increasing number of new seismic surveys and seismographic networks in South America, crustal thickness measurements are still scarce and irregularly sampled, reducing the resolution of crustal model maps. To overcome these challenges, a novel approach based on machine learning techniques is proposed, in order to explore higher resolution gravity datasets in the interpolation of crustal thickness measurement points obtained from previous seismic/seismological compilations. The algorithm used in this study is based on Gaussian processes prediction methods, which allowed inferring the depth of Moho to South America. The prediction error of the model obtained from the training and testing database was 3.48 km, which is compatible with the uncertainties derived from the H-k stacking analysis. The depth range varied from 69.8 km beneath the Andes to 4.3 km in oceanic regions. The average Moho depth for the South American Platform is 39.1 km, allowing a spatial correlation of deeper and shallower Moho regions with different types of continental basins. Compared to other models, the model resulting from this study presents fine-scale features highlighting the limits of the main tectonic domains and a good agreement with the suture zones. Overall, this study demonstrates the potential of applying machine learning tools in crustal-scale imaging using sparse datasets, providing new advances in Moho modeling of the South America, as well as new perspectives on its the history and tectonic evolution.

Application of machine learning and grey Taguchi technique for the development and optimization of a natural fiber hybrid reinforced polymer composite for aircraft body manufacture

Abstract The rapid expansion of the air transport industry raises significant sustainability concerns due to its substantial carbon emissions and contribution to global climate change. These emissions are closely linked to fuel consumption, which in turn is influenced by the weight of materials used in aircraft systems. This study extensively applied machine learning tools for the optimization of natural fiber-reinforced composite material production parameters for aircraft body application. The Taguchi optimization technique was used to study the effect of sisal fibers, glass fibers, fiber length, and NaOH treatment concentration on the performance of the materials. Multi-objective optimization methods like the grey relational analysis and genetic algorithm (using the MATLAB programming interface) were employed to obtain the best combination of the studied factors for low fuel consumption (low carbon emission) and high-reliability structural applications of aircraft. The models developed from regressional analysis had high accuracy of prediction, with R-Square values all >80%. Optimization of the grey relational analysis of the developed composite using the genetic algorithm showed the best process parameter to achieve low weight material for aircraft application to be 40% sisal, 5% glass fiber at 35 mm fiber length, and 5% NaOH concentration with grey relational analysis at the highest possible level, which is unity.

Machine Learning-Enabled Genome Mining and Bioactivity Prediction of Natural Products.

Natural products (NPs) produced by microorganisms and plants are a major source of drugs, herbicides, and fungicides. Thanks to recent advances in DNA sequencing, bioinformatics, and genome mining tools, a vast amount of data on NP biosynthesis has been generated over the years, which has been increasingly exploited to develop machine learning (ML) tools for NP discovery. In this review, we discuss the latest advances in developing and applying ML tools for exploring the potential NPs that can be encoded by genomic language and predicting the types of bioactivities of NPs. We also examine the technical challenges associated with the development and application of ML tools for NP research.

Deep learning self-image update procedure in a wavefront sensor based on the Talbot phenomenon under Gaussian illumination

A feature-based image update procedure using machine learning is proposed to use in preprocessing of self-images in a Talbot wavefront sensor. A variant of the recurrent neural network with backpropagation, which is one of most widely applied machine learning tools, is utilized to stabilize intensity distribution in self-images in the case of an optical beam with a Gaussian profile. Once well trained, the neural network can decrease pit image shifts caused by beam intensity distribution in the case of a cosine-like grating. It is shown that based on the proposed recurrent neural network, it is possible to decrease the shift error caused by the Gaussian beam up to nine times depending on the aberration order and value. Despite the decreasing shift error, the value of the error of the restored aberration coefficient does not decrease significantly because of the feature-vector extraction method. It is shown additionally that due to the spatial spectrum wideness, the proposed self-image procedure is not applicable to binary gratings on the example of binary gratings with square pits. Adequate simulations are implemented to demonstrate the effectiveness and accuracy of the proposed approach.

Recovering the lost potential of meadows to help mitigate challenges facing California’s forests and water supply

Groundwater systems in snow-dominated drainage areas supply cool baseflows that support instream and downstream users late into the dry season. Yet, these catchments are becoming rarer with climate change and anthropogenic pressures that threaten groundwater systems. Restoration of low-gradient meadows and streams can recover a catchment’s natural storage potential, especially in Mediterranean biomes such as the Sierra Nevada of California where summer groundwater recharge is scarce. The degradation of meadows due to intense human modification has decreased groundwater elevations and shifted wet meadow plant communities toward more xeric forest and shrub communities. We applied machine learning tools to find potential “lost meadows” that may no longer support high groundwater elevations or meadow vegetation but do exhibit basic geomorphic and climatic characteristics similar to existing meadows. The model reveals potential meadow habitat in the Sierra Nevada of nearly three times its current extent. We offer two conceptual applications of the model for incorporating meadows in watershed restoration planning. The first application focuses on strategically expanding wet meadows already associated with fuel breaks for increasing wildfire resistance. The second shows how meadow restoration in post-wildfire landscapes could increase capture of sediment from burned hillslopes where increased sediment storage would benefit water storage. Meadows are important habitats that have become degraded due to long-term overuse. Re-envisioning their potential extent shows that, with restoration, meadows could also serve as components of California’s multi-tiered efforts to manage pressing threats to its forests and water supply.

Forecasting carbon emissions future prices using the machine learning methods.

Due to the uncertainty surrounding the coupling and decoupling of natural gas, oil, and energy commodity futures prices, the current study seeks to investigate the interactions between energy commodity futures, oil price futures, and carbon emission futures from a forecasting perspective with implications for environmental sustainability. We employed daily data on natural gas futures prices, crude oil futures prices, carbon futures prices, and Dow Jones energy commodity futures prices from January 2018 to October 2021. For empirical analysis, we applied machine learning tools including traditional multiple linear regression (MLR), artificial neural network (ANN), support vector regression (SVR), and long short-term memory (LSTM). The machine learning analysis provides two key findings. First, the nonlinear frameworks outperform linear models in developing the relationships between future oil prices (crude oil and heating oil) and carbon emission futures prices. Second, the machine learning findings establish that when oil prices and natural gas prices display extreme movement, carbon emission futures prices react nonlinearly. Understanding the nonlinear dynamics of extreme movements can help policymakers design climate and environmental policies, as well as adjust natural gas and oil futures prices. We discuss important implications to sustainable development goals mainly SDG 7 and SDG 12.

Prediction of the Outcome of Pakistani Heart Failure Patients by Various Supervised Machine Learning Methods

Aim: To foresee the outcome of heart failure(HF) in Pakistani patients with potential predictors and through various machine learning (ML) methods. Study design:The secondary data of Pakistani patients is taken from the UCI repository in which a cross-sectional, analytical study was planned. Place and duration: This data was collected in April-December, 2015 at the Institute of Cardiology and Allied hospital Faisalabad-Pakistan. Methodology: The data set consisted of299 patients distributed among male (194) and female patients (105). Ages, serum sodium (SS), serum creatinine (SC), gender, smoking, high blood pressure (HBP), ejection fraction (EF), anemia, platelets, Creatinine Phosphokinase (CPK), and diabetes were considered as the potential predictors for predicting the outcome of HF.The data set was analyzed with the help of various machine learning (ML) predictive models including Logistic regression (LR), K-nearest neighbor (KNN), and Decision trees (DT). Results: The ages of the patients were within 60.833±11.894 years. Out of 299 patients, 129 were anemic, 105 had high blood pressure (HBP), and 96 had a smoking history. A statistical model was estimated by applying LR which assisted us in identifying the significant predictors. The sensitivity of the LRwas observed to be 92.1%, whereas 85.6% of the outcome of HF patients was correctly predicted by this model (LR) and DT achieved89.6% prediction accuracy. Conclusion:Since HF is a substantial reason for deaths in Pakistan. Therefore, the identification of its potential risk factors and its accurate prediction by some modern tools are highly demanded. This study applied ML tools for the said task and concluded that among all the fitted ML models, DT predicted the correct outcome for HF patients proficiently. Keywords: Heart failure, machine learning, logistic regression, k-nearest neighbor, decision trees

Improving medical term embeddings using UMLS Metathesaurus

BackgroundHealth providers create Electronic Health Records (EHRs) to describe the conditions and procedures used to treat their patients. Medical notes entered by medical staff in the form of free text are a particularly insightful component of EHRs. There is a great interest in applying machine learning tools on medical notes in numerous medical informatics applications. Learning vector representations, or embeddings, of terms in the notes, is an important pre-processing step in such applications. However, learning good embeddings is challenging because medical notes are rich in specialized terminology, and the number of available EHRs in practical applications is often very small.MethodsIn this paper, we propose a novel algorithm to learn embeddings of medical terms from a limited set of medical notes. The algorithm, called definition2vec, exploits external information in the form of medical term definitions. It is an extension of a skip-gram algorithm that incorporates textual definitions of medical terms provided by the Unified Medical Language System (UMLS) Metathesaurus.ResultsTo evaluate the proposed approach, we used a publicly available Medical Information Mart for Intensive Care (MIMIC-III) EHR data set. We performed quantitative and qualitative experiments to measure the usefulness of the learned embeddings. The experimental results show that definition2vec keeps the semantically similar medical terms together in the embedding vector space even when they are rare or unobserved in the corpus. We also demonstrate that learned vector embeddings are helpful in downstream medical informatics applications.ConclusionThis paper shows that medical term definitions can be helpful when learning embeddings of rare or previously unseen medical terms from a small corpus of specialized documents such as medical notes.

How much is the view from the window worth? Machine learning-driven hedonic pricing model of the real estate market

Understanding the customers’ perception of the value of constituent characteristics of a good is among the key questions in any pricing strategy. Hedonic pricing allows such an analysis and is frequently applied in economic fields. Although it is regarded as a benchmark in its original form, the availability of new data sources and the development of machine learning techniques created a space for further improvement. In this study, we propose a general framework for applying machine learning tools to enhance the hedonic pricing model in several directions. We do this, first, by adding image and text sources to conventional data and then by applying an advanced nonparametric prediction model. Lastly, we use model agnostic analysis to uncover new pricing factors and unravel complex relationships that could not be captured by conventional models.

A review of the application of machine learning in adult obesity studies

<abstract> <p>In obesity studies, several researchers have been applying machine learning tools to identify factors affecting human body weight. However, a proper review of strength, limitations and evaluation metrics of machine learning algorithms in obesity is lacking. This study reviews the status of application of machine learning algorithms in obesity studies and to identify strength and weaknesses of these methods. A scoping review of paper focusing on obesity was conducted. PubMed and Scopus databases were searched for the application of machine learning in obesity using different keywords. Only English papers in adult obesity between 2014 and 2019 were included. Also, only papers that focused on controllable factors (e.g., nutrition intake, dietary pattern and/or physical activity) were reviewed in depth. Papers on genetic or childhood obesity were excluded. Twenty reviewed papers used machine learning algorithms to identify the relationship between the contributing factors and obesity. Regression algorithms were widely applied. Other algorithms such as neural network, random forest and deep learning were less exploited. Limitations regarding data priori assumptions, overfitting and hyperparameter optimization were discussed. Performance metrics and validation techniques were identified. Machine learning applications are positively impacting obesity research. The nature and objective of a study and available data are key factors to consider in selecting the appropriate algorithms. The future research direction is to further explore and take advantage of the modern methods, i.e., neural network and deep learning, in obesity studies.</p> </abstract>

Coupling 1D xRAGE simulations with machine learning for graded inner shell design optimization in double shell capsules

Advances in machine learning provide the ability to leverage data from expensive simulations of high-energy-density experiments to significantly cut down on computational time and costs associated with the search for optimal target designs. This study presents an application of cutting-edge Bayesian optimization methods to the one-dimensional (1D) design optimization of double shell graded layer targets for inertial confinement fusion experiments. This investigation attempts to reduce hydrodynamic instabilities while retaining high yields for future NIF experiments. Machine learning methods can use predictive physics simulations to identify graded layer designs from within the vast design space that demonstrate high predicted performance, including novel designs with high uncertainty in performance that may hold unexpected promise. By applying machine learning tools to the simulation design, we map the trade-off between 1D yield and instability, specifically isolating parameter ranges, which maintain high performance while showing significantly improved Rayleigh–Taylor stability over the point design. The groundwork laid in this study will be a useful design tool for future NIF experiments with graded layer targets.

Evaluating the Potential of Applying Machine Learning Tools to Metabolic Pathway Optimization

Successful engineering of a microbial host for efficient production of a target product from a given substrate can be viewed as an extensive optimization task. Such a task involves the selection of high activity enzymes as well as their gene expression regulatory control elements (i.e., promoters and ribosome binding sites). Finally, there is also the need to tune expression of multiple genes along a heterologous pathway to relieve constraints from rate-limiting step and help reduce metabolic burden on cells from unnecessary over-expression of high activity enzymes. While the aforementioned tasks could be performed through combinatorial experiments, such an approach incurs significant cost, time and effort, which is a handicap that can be relieved by application of modern machine learning tools. Such tools could attempt to predict high activity enzymes from sequence, but they are currently most usefully applied in classifying strong promoters from weaker ones as well as combinatorial tuning of expression of multiple genes. This perspective reviews the application of machine learning tools to aid metabolic pathway optimization through identifying challenges in metabolic engineering that could be overcome with the help of machine learning tools.

Deep neural networks using a single neuron: folded-in-time architecture using feedback-modulated delay loops

Deep neural networks are among the most widely applied machine learning tools showing outstanding performance in a broad range of tasks. We present a method for folding a deep neural network of arbitrary size into a single neuron with multiple time-delayed feedback loops. This single-neuron deep neural network comprises only a single nonlinearity and appropriately adjusted modulations of the feedback signals. The network states emerge in time as a temporal unfolding of the neuron’s dynamics. By adjusting the feedback-modulation within the loops, we adapt the network’s connection weights. These connection weights are determined via a back-propagation algorithm, where both the delay-induced and local network connections must be taken into account. Our approach can fully represent standard Deep Neural Networks (DNN), encompasses sparse DNNs, and extends the DNN concept toward dynamical systems implementations. The new method, which we call Folded-in-time DNN (Fit-DNN), exhibits promising performance in a set of benchmark tasks.

Leveraging low-cost air quality sensors and machine learning techniques for air pollution assessment and prediction in urban Ghana

BACKGROUND AND AIM: In recent times, there has been a proliferation of low-cost air quality sensors (LCS). In Africa, where air quality monitoring systems are very weak, LCS offer a great opportunity for bridging the huge air pollution data gaps in countries. METHODS: The Ghana Urban Air Quality Project (GHAir) initiated in May 2019 and presently covering five cities of Ghana leverages PurpleAir sensors, clarity nodes, and Real-Time Multi-Pollutant sensors (RAMP) to monitor air pollution in these settlements for epidemiologic research whilst also applying machine learning tools to gleaned insight into the air quality data generated for community advocacy to improve air quality, and influence air pollution control policies. RESULTS:In this presentation, I would be providing data on the levels and trends of PM2.5, NO2, SO2 and ground level ozone in the city of Accra especially and other metropolitan areas for the period May 2019 to July 2021. I would also present results from prediction models depicting future air pollution situation in Accra and outline measures that can be help to mitigate the situation. CONCLUSIONS:The GHAir project has helped to bridge the air pollution data gap in urban Ghana and should be supported by regional and global clean air interest groups to scale up the project and to share the lessons learnt in other African countries KEYWORDS: Air pollution, Data science, Ghana, Low cost sensors

Machine learning for cardiology.

This paper reviews recent cardiology literature and reports how artificial intelligence tools (specifically, machine learning techniques) are being used by physicians in the field. Each technique is introduced with enough details to allow the understanding of how it works and its intent, but without delving into details that do not add immediate benefits and require expertise in the field. We specifically focus on the principal Machine learning based risk scores used in cardiovascular research. After introducing them and summarizing their assumptions and biases, we discuss their merits and shortcomings. We report on how frequently they are adopted in the field and suggest why this is the case based on our expertise in machine learning. We complete the analysis by reviewing how corresponding statistical approaches compare with them. Finally, we discuss the main open issues in applying machine learning tools to cardiology tasks, also drafting possible future directions. Despite the growing interest in these tools, we argue that there are many still underutilized techniques: while neural networks are slowly being incorporated in cardiovascular research, other important techniques such as semi-supervised learning and federated learning are still underutilized. The former would allow practitioners to harness the information contained in large datasets that are only partially labeled, while the latter would foster collaboration between institutions allowing building larger and better models.

Personalized Survival Prediction of Patients With Acute Myeloblastic Leukemia Using Gene Expression Profiling.

Acute Myeloid Leukemia (AML) is a heterogeneous neoplasm characterized by cytogenetic and molecular alterations that drive patient prognosis. Currently established risk stratification guidelines show a moderate predictive accuracy, and newer tools that integrate multiple molecular variables have proven to provide better results. In this report, we aimed to create a new machine learning model of AML survival using gene expression data. We used gene expression data from two publicly available cohorts in order to create and validate a random forest predictor of survival, which we named ST-123. The most important variables in the model were age and the expression of KDM5B and LAPTM4B, two genes previously associated with the biology and prognostication of myeloid neoplasms. This classifier achieved high concordance indexes in the training and validation sets (0.7228 and 0.6988, respectively), and predictions were particularly accurate in patients at the highest risk of death. Additionally, ST-123 provided significant prognostic improvements in patients with high-risk mutations. Our results indicate that survival of patients with AML can be predicted to a great extent by applying machine learning tools to transcriptomic data, and that such predictions are particularly precise among patients with high-risk mutations.

A modified NEFCLASS classifier with enhanced accuracy-interpretability trade-off for datasets with skewed feature values

The accuracy-transparency trade-off is one of the most notable challenges when applying machine learning tools in the medical domain. Nefclass is a popular neuro-fuzzy classifier in medical diagnosis systems. Nefclass performs increasingly poorly as the data skewness increases. This paper presents a combined approach to improve the classification accuracy and interpretability of the Nefclass classifier, when feature values of the training and testing datasets exhibit positive skewness. The proposed model consists of two steps. Firstly, a modified Nefclass classifier embedded with a choice of two alternative discretization methods, MME and CAIM is implemented. Secondly, we devised a new rule pruning method based on the Habermans' adjusted residual to reduce the size of the resulting ruleset. This rule-pruning method improves the interpretability of Nefclass without significant accuracy deterioration. Moreover, a hybrid approach combining the two approaches provides a considerable improvement in classification accuracy and transparency of Nefclass on skewed data.

Machine learning applications in nonlife insurance

AbstractThe literature on analytical applications in insurance tends to be either very general or rather technical, which may hold back the adoption of new important tools by industrial practitioners. Our goal is to stress that machine learning (ML) algorithms will play a significant role in the insurance industry in the near future and thus to encourage practitioners to learn and apply these techniques. After discussing the increasing relevance of data for nonlife insurance and briefly reviewing the major impact of digital technology on this business, we restrict our discussion to technical analytical applications and indicate where ML algorithms can add most value. We present two real examples: first a comparison of retention models for household insurance and then a dynamic pricing problem for online motor insurance. Both applications illustrate the advantages but also some of the difficulties of applying ML tools in practice. Finally, we mention some challenges posed by the use of ML in the industry and formulate a few recommendations for successful applications in insurance. This article is neither a tutorial nor an exhaustive review of technical ML applications in nonlife insurance. However, references for additional learning materials are provided.

P91 Bringing patients into the referral process: an evaluation of pre-appointment questionnaires in a rheumatology new patient pathway

Abstract Background Referrals from primary care to rheumatology services have increased significantly in recent years. Many of these patients have non-inflammatory conditions, usually with pain, and may require specialist treatment such as physiotherapy or psychological intervention. If patients could be effectively triaged at the point of referral, clinical services could be better aligned to patients’ needs, but referrals do not always contain enough information to triage patients effectively. Validated patient-completed questionnaires are available which are designed to aid patient identification, for instance for fibromyalgia and inflammatory back pain. We designed an electronic questionnaire for patients to complete prior to their first attendance to inform the clinical consultation and to explore whether they could be more appropriately triaged to specialist clinics. Methods We constructed a questionnaire consisting of the modified ACR diagnostic questionnaire for fibromyalgia, the ASAS screening questionnaire for inflammatory back pain, a validated questionnaire for identification of patients with inflammatory arthritis, PHQ8 (depression) and GAD 7 (anxiety). The questionnaire contained conditional branching to ensure that only relevant questions were answered. A user-centred design approach was taken to optimise patient experience and response rates. A link to the questionnaire was sent via SMS to unselect new patients shortly after receipt of a referral from their GP. Completed questionnaires were imported into the hospital record. Results Initial response rates were around 35%. Through a process of continuous iteration informed by user feedback this increased to 80% over a 4 months period. Key factors in optimisation were the appearance of the text (displaying from ‘NHS’ rather than an apparently random number), the language which was used to maximise patients perception of the value of completion, the use of patients’ first names in the SMS and the content and timing of automated reminder messages. Over a 4 month after optimisation 664 patients completed the questionnaire. Of these 38% and 25% scored in the moderate-severe ranges for depression and anxiety respectively. 74% of patients met the paper criteria for fibromyalgia (which is not diagnostic in itself prior to clinical assessment). Of those patients scoring in the moderate-severe ranges for depression and anxiety, 92% and 88% respectively met the questionnaire criteria for fibromyalgia. Conclusion This work suggests that collecting information prior to the first appointment can add value to the process. In our department screening tools for psychological distress, widespread pain and somatisation are used to inform the consultation. We are currently analysing this data against the patients’ final diagnoses. Combination of screening tools may enable patients to be stratified with enough predictive value into specific pathways for inflammatory and non-inflammatory conditions, some of which could be community-based. We will be applying machine learning tools to determine the optimum predictive value of these questionnaires in combination. Disclosures W. Chen None. M. Thrower None. T. Garrood None.