Machine Learning for Materials Research and Development

Perspective on machine learning in energy material discovery

Carotid artery stenting (CAS) carries important perioperative risks. Outcome prediction tools may help guide clinical decision-making but remain limited. We developed machine learning (ML) algorithms that predict 30-day outcomes following transfemoral CAS. The National Surgical Quality Improvement Program (NSQIP) targeted vascular database was used to identify patients who underwent transfemoral CAS between 2011 and 2021. Input features included 36 preoperative demographic/clinical variables. The primary outcome was a 30-day major adverse cardiovascular event (MACE; composite of stroke, myocardial infarction [MI], or death). The secondary outcomes were 30-day stroke, MI, death, carotid-related morbidity, other morbidity, non-home discharge, and unplanned readmission. Our data were split into training (70%) and test (30%) sets. Using 10-fold cross-validation, we trained six ML models using preoperative features with logistic regression as the baseline comparator. The primary model evaluation metric was area under the receiver operating characteristic curve (AUROC). Model robustness was evaluated with calibration plot and Brier score. Variable importance scores were calculated to determine the top 10 predictive features. Performance was assessed on subgroups based on age, sex, race, ethnicity, symptom status, stent type, and urgency. Overall, 2093 patients underwent CAS during the study period. Thirty-day MACE occurred in 130 (6.2%) patients. The best-performing prediction model for 30-day MACE was XGBoost, achieving an AUROC (95% CI) of 0.93 (0.92-0.94). In comparison, logistic regression had an AUROC (95% CI) of 0.67 (0.65-0.68), and existing tools in the literature demonstrate AUROCs ranging from 0.58 to 0.74. For secondary outcomes, XGBoost achieved AUROCs between 0.86 and 0.97. The calibration plot showed good agreement between predicted and observed event probabilities with a Brier score of 0.02. The top three predictive features in our algorithm were (1) symptomatic carotid stenosis, (2) age, and (3) American Society of Anesthesiologists classification. Model performance remained robust on all subgroup analyses of specific demographic and clinical populations. Our ML models accurately predict 30-day outcomes following transfemoral CAS using preoperative data. They have the potential for important utility in guiding risk-mitigation strategies for patients being considered for CAS to improve outcomes.Clinical ImpactTransfemoral carotid artery stenting (CAS) carries important perioperative risks. Outcome prediction tools may help guide clinical decision-making but remain limited. Using data from the National Surgical Quality Improvement Program (NSQIP) targeted vascular database, we developed machine learning (ML) models that accurately predict 30-day outcomes following transfemoral CAS using preoperative data, outperforming logistic regression and existing tools in the literature. The models were well-calibrated and remained robust across demographic and clinical subpopulations. These ML algorithms have the potential for important utility in guiding risk-mitigation strategies for patients being considered for transfemoral CAS to improve outcomes.

Machine learning for sustainable development and applications of biomass and biomass-derived carbonaceous materials in water and agricultural systems: A review

With its unique advantages in artificial intelligence, data analysis, interpolation and numerical extrapolation, etc. ML has recently been quickly developed for the discovery of advanced energy materials. In particular, many algorithms have been developed to predict material properties. Herein, we first introduce the ML algorithms used in material science and the structure of each algorithm. Then we examine the algorithms that have been used recently in functional materials, especially in solar cells, batteries, and phase-change materials. Finally, advantages and disadvantages of each algorithm are compared to aid readers in choosing a suitable algorithm for specific applications.

The identification of suitable alloy compositions for the formation of bulk metallic glasses (BMGs) is a key challenge in materials science. In this study, we developed machine learning (ML) models to predict the critical casting diameter (Dmax) of Fe-based BMGs, enabling rapid assessment of glass-forming ability (GFA) using composition-based and calculated thermophysical features. Three datasets were constructed: one based on alloy molar fractions, one using thermophysical quantities calculated via the CALPHAD method, and another utilizing Magpie-derived features. The performance of various ML models was evaluated, including support vector machines (SVM), XGBoost, and ensemble methods. Models trained on thermophysical features outperformed those using only molar fractions, with XGBoost and SVM models achieving test R2 scores of up to 0.63 and 0.60, respectively. Magpie features yielded similar results but required a larger feature set. To enhance predictive accuracy, we explored data augmentation using the PADRE method and a modified version (PADRE-2). While PADRE-2 demonstrated slight improvements and reduced data redundancy, the overall performance gains were limited. The best-performing model was an ensemble combining SVM and XGBoost models trained on thermophysical and Magpie features, achieving an R2 score of 0.69 and MAE of 0.69, comparable to published results obtained from larger datasets. However, predictions for high Dmax values remain challenging, highlighting the need for further refinement. This study underscores the potential of leveraging thermophysical features and advanced ML techniques for GFA prediction and the design of new Fe-based BMGs.

Applications of Machine Learning in Solid Oral Dosage Form Development.

Materials are not only the foundation of the national economy, but also the carrier of high technology. It has become a research hotspot in the world to overcome the conventional methods and apply new methods to accelerate the development of new materials. Propelled by the great success in other fields, data-driven informatics methods begin to emerge as a new technique in material science. Machine learning, as a representative of data-driven methods, has received extensive attention in various fields. Machine learning is an interdisciplinary science that combines computer science, statistics, computational mathematics and engineering. In the field of materials science, machine learning methods show faster calculation speed and higher prediction accuracy compared with conventional theoretical computational simulations based on solving physical or chemical fundamental equations. Machine learning is an effective addition to the existing theoretical calculation methods and significantly increases the efficiency of materials computational simulation work. Furthermore, it also works for some systems or problems that the traditional theoretical calculation methods fail to solve. This approach could also enable targeted material design and development. This review would provide a brief overview on the fundamentals of machine learning, several typical algorithms in machine learning and the applications in materials science, and discuss the future challenges in this field.

Cities are getting smarter every day. Municipalities are increasingly using information and communication technologies (ICT) to enrich and enhance city life, which is paramount in planning the cities of the future. This calls for connected and automated cars in the future. This emphasizes the need for efficient and fast acting controls algorithms in Engine Management System (EMS) to achieve enhanced dynamic response of the connected mobility. Artificial Intelligence and Machine learning in development of control algorithms is of paramount need. Advancements in internal combustion engine technologies have increased a need of complex features. Effective control of these features is the need for the day. The optimum solution of these features via parameter optimization needs to be effectively derived for entire engine operating range in steady and transient operation. This is very critical in case of urban mobility where synchronized and harmonized management is the need of the hour. Today developers apply supervised learning to powertrain calibration, missing physics and signature, emission control and virtual sensor development.This paper presents the implementation of machine learning algorithms for predictive models via optimal set of parameters for various powertrain features for synchronized mobility. Predictive models amalgamated with Artificial Intelligence will accurately determine the response of a system. Machine learning techniques is used to create a predictive model when no knowledge of the system is known and difficult to determine. The paper also presents the rethinking of conventional process by introducing Machine learning in all the phases mentioned: Concept → Development → SOP → series volume production.

Tiny Machine Learning (TML) is a novel research area aiming at designing and developing Machine Learning (ML) techniques meant to be executed on Embedded Systems and Internet-of-Things (IoT) units. Such techniques, which take into account the constraints on computation, memory, and energy characterizing the hardware platform they operate on, exploit approximation and pruning mechanisms to reduce the computational load and the memory demand of Machine and Deep Learning (DL) algorithms.Despite the advancement of the research, TML solutions present in the literature assume that Embedded Systems and IoT units support only the inference of ML and DL algorithms, whereas their training is confined to more-powerful computing units (due to larger computational load and memory demand). This also prevents such pervasive devices from being able to learn in an incremental way directly from the field to improve the accuracy over time or to adapt to new working conditions.The aim of this paper is to address such an open challenge by introducing an incremental algorithm based on transfer learning and k-nearest neighbor to support the on-device learning (and not only the inference) of ML and DL solutions on embedded systems and IoT units. Moreover, the proposed solution is general and can be applied to different application scenarios. Experimental results on image/audio benchmarks and two off-the-shelf hardware platforms show the feasibility and effectiveness of the proposed solution.

The impact of manufacturing technologies in the electronics field on the development of materials and components is evaluated. As an example, it is shown that the advanced manufacturing technology for ultrafine particles has clearly had a substantial influence on the development of new materials and components. Also considered is the impact of the development of materials and components on manufacturing technologies. It is concluded that since the development of materials and components would not be possible without the support of manufacturing technologies, their development must be conducted more tightly in tandem with manufacturing technologies. >

The book will focus on the applications of machine learning for sustainable development. Machine learning (ML) is an emerging technique whose diffusion and adoption in various sectors (such as energy, agriculture, internet of things, infrastructure) will be of enormous benefit. The state of the art of machine learning models is most useful for forecasting and prediction of various sectors for sustainable development.

Who is in control of artificial intelligence and can we trust them?

In order to mechanically predict audiovisual quality in interactive multimedia services, we have developed machine learning--based no-reference parametric models. We have compared Decision Trees--based ensemble methods, Genetic Programming and Deep Learning models that have one and more hidden layers. We have used the Institut national de la recherche scientifique (INRS) audiovisual quality dataset specifically designed to include ranges of parameters and degradations typically seen in real-time communications. Decision Trees--based ensemble methods have outperformed both Deep Learning-- and Genetic Programming--based models in terms of Root-Mean-Square Error (RMSE) and Pearson correlation values. We have also trained and developed models on various publicly available datasets and have compared our results with those of these original models. Our studies show that Random Forests--based prediction models achieve high accuracy for both the INRS audiovisual quality dataset and other publicly available comparable datasets.

Machine Learning for Drug Discovery is designed to suit the needs of graduate students, advanced undergraduates, chemists or biologists otherwise new to this research domain with minimal previous exposure to Machine Learning (ML) methods, or computational scientists with minimal exposure to medicinal chemistry. The e-book covers basic algorithmic theory, data representation methods, and generative modeling at a high level. The authors spotlight antibiotic discovery as a case study in ML for drug development and discuss diverse applications in drug-likeness prediction, antimicrobial resistance, and areas for future inquiry. For a more dynamic learning experience, open-source code demonstrations in Python are included.

This article presents a comprehensive literature study showing that, starting from the emergence of the species of homo sapiens, the progress of human civilization is strongly dependent on the development of materials, and over time this is mainly development of engineering materials and the accompanying increase in productive forces. There is no close correlation between changes, and especially between the development of the brain and technologies sequentially controlled by humans. The materials science appeared as an independent branch of knowledge only in the late 1950s. Technical aspects of the product launch on the market relate to several technical aspects; engineering design is a significant conceptual phase of this activity, and within it, the material design. The expected functional properties of the product will be assured only if the right expected material is used and is produced in a suitably selected expected technological process that will provide both the expected shape and other geometric features of the product, including assembly and the expected structure of the material, ensuring the expected mechanical, physical, and chemical properties of the material, the use of which ensures the expected application of the product. The given rule defines the paradigm of contemporary materials science and engineering. Without the use of engineering materials and without the development of manufacturing processes, it is impossible to manufacture any product and make it available to consumers. The material design process in history has gone through a long period of changes. Initially, for millions of years and almost a century ago, materials were selected based on the trial and error method, which is stage Materials 1.0. Currently, about 80 % of work in the field of engineering materials development and material design is carried out in accordance with the Materials 2.0 protocol. The Materials 2.0 protocol includes more systematic material research, ranging from conceptualization, systematic laboratory experiments to verify the idea, prototyping in the laboratory and real conditions, and testing and validating prototypes and life cycle assessment to use the results of research in product production. Materials 3.0 use computational materials science materials, and materials are computationally designed with a target functionality. Only the idea of Materials 4.0 overcomes human limitations in applying existing knowledge about the theory of materials, processing, and properties through the use of cyber-physical space. In manufacturing processes generally after the era of water and steam and mass production based on the division of labor with the use of electricity, and then the use of electronics and information technology for the automation of production processes, there is a dynamic use of cyber-physical systems, the Internet of objects, machine learning, artificial intelligence, and virtual reality, currently guaranteeing the production progress at the stage of the Industry 4.0 industrial revolution. So far, nine technologies have been designated as determining the change in industrial production at the Industry 4.0 stage. According to the authors of this article, it is necessary to augment this list by manufacturing processes and engineering materials as well as living and bioengineering machines; therefore, in total, there will be 12 technologies determining the changes in production in the Industry 4.0 stage. Omitting these issues would make it impossible to manufacture any products available on the market, and the idea of Industry 4.0 presented would be incomplete. The importance of carbon-based materials is also presented, and several results of our own research on various materials are presented, with an indication of application possibilities.