scholarly journals Machine Learning in Model-free Mechanical Property Imaging: Novel Integration of Physics With the Constrained Optimization Process

2021 ◽  
Vol 9 ◽  
Author(s):  
Cameron Hoerig ◽  
Jamshid Ghaboussi ◽  
Yiliang Wang ◽  
Michael F. Insana
Keyword(s):  
Machine Learning ◽  
Mechanical Properties ◽  
Measurement Data ◽  
Optimization Methods ◽  
Patient Specific ◽  
Learning Approaches ◽  
Element Analysis ◽  
Physical Constraints ◽  
Model Free ◽  
Force Displacement

The Autoprogressive Method (AutoP) is a fundamentally different approach to solving the inverse problem in quasi-static ultrasonic elastography (QUSE). By exploiting the nonlinear adaptability of artificial neural networks and physical constraints imposed through finite element analysis, AutoP is able to build patient specific soft-computational material models from a relatively sparse set of force-displacement measurement data. Physics-guided, data-driven models offer a new path to the discovery of mechanical properties most effective for diagnostic imaging. AutoP was originally applied to modeling mechanical properties of materials in geotechnical and civil engineering applications. The method was later adapted to reconstructing maps of linear-elastic material properties for cancer imaging applications. Previous articles describing AutoP focused on high-level concepts to explain the mechanisms driving the training process. In this review, we focus on AutoP as applied to QUSE to present a more thorough explanation of the ways in which the method fundamentally differs from classic model-based and other machine learning approaches. We build intuition for the method through analogy to conventional optimization methods and explore how maps of stresses and strains are extracted from force-displacement measurements in a model-free way. In addition, we discuss a physics-based regularization term unique to AutoP that illuminates the comparison to typical optimization procedures. The insights gained from our hybrid inverse method will hopefully inspire others to explore combinations of rigorous mathematical techniques and conservation principles with the power of machine learning to solve difficult inverse problems.

scholarly journals Inference of an Integrative, Executable Network for Rheumatoid Arthritis Combining Data-Driven Machine Learning Approaches and a State-of-the-Art Mechanistic Disease Map

2021 ◽  
Vol 11 (8) ◽  
pp. 785
Author(s):  
Quentin Miagoux ◽  
Vidisha Singh ◽  
Dereck de Mézquita ◽  
Valerie Chaudru ◽  
Mohamed Elati ◽  
...  
Keyword(s):  
Rheumatoid Arthritis ◽  
Machine Learning ◽  
State Of The Art ◽  
Biological Information ◽  
Response To Treatment ◽  
Patient Specific ◽  
Learning Approaches ◽  
Data Types ◽  
Disease Heterogeneity ◽  
Combining Data

Rheumatoid arthritis (RA) is a multifactorial, complex autoimmune disease that involves various genetic, environmental, and epigenetic factors. Systems biology approaches provide the means to study complex diseases by integrating different layers of biological information. Combining multiple data types can help compensate for missing or conflicting information and limit the possibility of false positives. In this work, we aim to unravel mechanisms governing the regulation of key transcription factors in RA and derive patient-specific models to gain more insights into the disease heterogeneity and the response to treatment. We first use publicly available transcriptomic datasets (peripheral blood) relative to RA and machine learning to create an RA-specific transcription factor (TF) co-regulatory network. The TF cooperativity network is subsequently enriched in signalling cascades and upstream regulators using a state-of-the-art, RA-specific molecular map. Then, the integrative network is used as a template to analyse patients’ data regarding their response to anti-TNF treatment and identify master regulators and upstream cascades affected by the treatment. Finally, we use the Boolean formalism to simulate in silico subparts of the integrated network and identify combinations and conditions that can switch on or off the identified TFs, mimicking the effects of single and combined perturbations.

scholarly journals Application of Machine Learning to Bending Processes and Material Identification

Metals ◽  
2021 ◽  
Vol 11 (9) ◽  
pp. 1418
Author(s):  
Daniel J. Cruz ◽  
Manuel R. Barbosa ◽  
Abel D. Santos ◽  
Sara S. Miranda ◽  
Rui L. Amaral
Keyword(s):  
Machine Learning ◽  
Sheet Metal ◽  
Bending Test ◽  
Model Parameters ◽  
Learning Approaches ◽  
Additional Information ◽  
The Many ◽  
And Control ◽  
Force Displacement ◽  
Springback Angle

The increasing availability of data, which becomes a continually increasing trend in multiple fields of application, has given machine learning approaches a renewed interest in recent years. Accordingly, manufacturing processes and sheet metal forming follow such directions, having in mind the efficiency and control of the many parameters involved, in processing and material characterization. In this article, two applications are considered to explore the capability of machine learning modeling through shallow artificial neural networks (ANN). One consists of developing an ANN to identify the constitutive model parameters of a material using the force–displacement curves obtained with a standard bending test. The second one concentrates on the springback problem in sheet metal press-brake air bending, with the objective of predicting the punch displacement required to attain a desired bending angle, including additional information of the springback angle. The required data for designing the ANN solutions are collected from numerical simulation using finite element methodology (FEM), which in turn was validated by experiments.

scholarly journals Comparison of Empirical Mode Decomposition, Wavelets, and Different Machine Learning Approaches for Patient-Specific Seizure Detection Using Signal-Derived Empirical Dictionary Approach

2021 ◽  
Vol 3 ◽  
Author(s):  
Muhammad Kaleem ◽  
Aziz Guergachi ◽  
Sridhar Krishnan
Keyword(s):  
Machine Learning ◽  
Seizure Detection ◽  
Reconstruction Error ◽  
Data Driven ◽  
Patient Specific ◽  
Learning Approaches ◽  
Mode Decomposition ◽  
Eeg Data ◽  
Classifier Training

Analysis of long-term multichannel EEG signals for automatic seizure detection is an active area of research that has seen application of methods from different domains of signal processing and machine learning. The majority of approaches developed in this context consist of extraction of hand-crafted features that are used to train a classifier for eventual seizure detection. Approaches that are data-driven, do not use hand-crafted features, and use small amounts of patients' historical EEG data for classifier training are few in number. The approach presented in this paper falls in the latter category, and is based on a signal-derived empirical dictionary approach, which utilizes empirical mode decomposition (EMD) and discrete wavelet transform (DWT) based dictionaries learned using a framework inspired by traditional methods of dictionary learning. Three features associated with traditional dictionary learning approaches, namely projection coefficients, coefficient vector and reconstruction error, are extracted from both EMD and DWT based dictionaries for automated seizure detection. This is the first time these features have been applied for automatic seizure detection using an empirical dictionary approach. Small amounts of patients' historical multi-channel EEG data are used for classifier training, and multiple classifiers are used for seizure detection using newer data. In addition, the seizure detection results are validated using 5-fold cross-validation to rule out any bias in the results. The CHB-MIT benchmark database containing long-term EEG recordings of pediatric patients is used for validation of the approach, and seizure detection performance comparable to the state-of-the-art is obtained. Seizure detection is performed using five classifiers, thereby allowing a comparison of the dictionary approaches, features extracted, and classifiers used. The best seizure detection performance is obtained using EMD based dictionary and reconstruction error feature and support vector machine classifier, with accuracy, sensitivity and specificity values of 88.2, 90.3, and 88.1%, respectively. Comparison is also made with other recent studies using the same database. The methodology presented in this paper is shown to be computationally efficient and robust for patient-specific automatic seizure detection. A data-driven methodology utilizing a small amount of patients' historical data is hence demonstrated as a practical solution for automatic seizure detection.

scholarly journals How to gauge investor behavior? A comparison of online investor sentiment measures

2021 ◽  
Author(s):  
Daniele Ballinari ◽  
Simon Behrendt
Keyword(s):  
Machine Learning ◽  
Abnormal Returns ◽  
Investor Sentiment ◽  
Investor Behavior ◽  
Learning Approaches ◽  
Financial Applications ◽  
Model Free ◽  
Retail Investors ◽  
Regression Framework ◽  
Social Media Platforms

AbstractGiven the increasing interest in and the growing number of publicly available methods to estimate investor sentiment from social media platforms, researchers and practitioners alike are facing one crucial question – which is best to gauge investor sentiment? We compare the performance of daily investor sentiment measures estimated from Twitter and StockTwits short messages by publicly available dictionary and machine learning based methods for a large sample of stocks. To determine their relevance for financial applications, these investor sentiment measures are compared by their effects on the cross-section of stocks (i) within a Fama and MacBeth (J Polit Econ 81:607–636, 1973) regression framework applied to a measure of retail investors’ order imbalances and (ii) by their ability to forecast abnormal returns in a model-free portfolio sorting exercise. Interestingly, we find that investor sentiment measures based on finance-specific dictionaries do not only have a greater impact on retail investors’ order imbalances than measures based on machine learning approaches, but also perform very well compared to the latter in our asset pricing application.

scholarly journals Automated and Data-driven Computational Design of Patient-Specific Biomechanical Interfaces

2016 ◽  
Author(s):  
Kevin Mattheus Moerman ◽  
Dana Solav ◽  
David Sengeh ◽  
Hugh Herr
Keyword(s):  
Mechanical Properties ◽  
Load Transfer ◽  
Computational Design ◽  
Data Driven ◽  
Patient Specific ◽  
Element Analysis ◽  
Computational Framework ◽  
Quantitative Evidence ◽  
Non Invasive ◽  
Tissue Region

Biomechanical interfaces are mechanical structures that form the connection between a device and a tissue region, and, through appropriate load transfer, aim to minimize tissue discomfort and injury. A patient-specific and data-driven computational framework for the automated design of biomechanical interfaces is presented here. Optimization of the design of biomechanical interfaces is complex since it is affected by the interplay of the geometry and mechanical properties of both the tissue and the interface. The proposed framework is presented for the application of transtibial amputee prostheses where the interface is formed by a prosthetic liner and socket. Conventional socket design and manufacturing is largely artisan, non-standard, and insufficiently data-driven, leading to discrepancies between the quality of sockets produced by different prosthetists. Furthermore, current prosthetic liners are often not patient-specific. The proposed framework involves: A) non-invasive imaging to record patient geometry, B) indentation to assess tissue mechanical properties, C) data-driven and automated creation of patient-specific designs, D) patient-specific finite element analysis (FEA) and design evaluation, and finally E) computer aided manufacturing. Uniquely, the FEA procedure controls both the design and mechanical properties of the devices, and simulates, not only the loading during use, but also the pre-load induced by the donning of both the liner and the socket independently. Through FEA evaluation, detailed information on internal and external tissue loading, which are directly responsible for discomfort and injury, are available. Further, these provide quantitative evidence on the implications of design choices, e.g. : 1) alterations in the design can be used to locally enhance or reduce tissue loading, 2) compliant features can aid in relieving local surface pressure. The proposed methods form a patient-specific, data-driven and repeatable design framework for biomechanical interfaces, and by enabling FEA-based optimization reduces the requirement for repeated patient involvement in the currently manual and iterative design process.

scholarly journals Einstein-Roscoe Regression for the Slag Viscosity Prediction Problem in Steelmaking

2022 ◽  
Author(s):  
Hiroto Saigo ◽  
K.C. Dukka Bahadur ◽  
Noritaka Saito
Keyword(s):  
Machine Learning ◽  
High Temperature ◽  
Slag System ◽  
Natural Sciences ◽  
Slag Viscosity ◽  
Learning Approaches ◽  
Prediction Problem ◽  
Physical Constraints ◽  
Regression Parameters ◽  
Viscosity Prediction

Abstract In classical machine learning, regressors are trained without attempting to gain insight into the mechanism connecting inputs and outputs. Natural sciences, however, are interested in finding a robust interpretable function for the target phenomenon, that can return predictions even outside of the training domains. This paper focuses on viscosity prediction problem in steelmaking, and proposes Einstein-Roscoe regression (ERR), which learns the coefficients of the Einstein-Roscoe equation, and is able to extrapolate to unseen domains. Besides, it is often the case in the natural sciences that some measurements are much more expensive than the others due to physical constraints. To this end, we employ a transfer learning framework based on Gaussian process, which allows us to estimate the regression parameters using the auxiliary measurements available in a reasonable cost. In experiments using the viscosity measurements in high temperature slag system, ERR is compared favorably with various machine learning approaches in interpolation settings, while outperformed all of them in extrapolation settings. Furthermore, after estimating parameters using the auxiliary dataset obtained at room temperature, increase in accuracy is observed in the high temperature dataset, which corroborates the effectiveness of the proposed approach.

scholarly journals Design and Mechanical Properties of a New Clamp-Type Carbon Fiber Materials Tension Anchoring System

2021 ◽  
Vol 2101 (1) ◽  
pp. 012060
Author(s):  
Xianwei Wang ◽  
Bo Liu ◽  
Shengyong Hu ◽  
Longsheng Bao
Keyword(s):  
Mechanical Properties ◽  
Carbon Fiber ◽  
Measurement Data ◽  
Fiber Material ◽  
Element Analysis ◽  
Anchor System ◽  
Carbon Fiber Material ◽  
Inclination Angles ◽  
Anchoring System ◽  
Fiber Materials

Abstract Carbon fiber materials are widely used in bridge reinforcement techniques, while conventional carbon fiber material tensile anchoring equipment produces a large prestressed loss. This paper analyzes the deficiencies of existing tensile anchoring systems at home and abroad, summarizing the cause of prestressed losses, and combining with existing anchoring systems, a new type of clamp type carbon fiber cloth tension anchoring system is proposed. The amount of deformation of the anchoring system is reduced by about 20%, which in turn reduces the system prestress loss caused by the system deformation. The ABAQUS finite element analysis software is used to numerically simulate the thickness of the tension anchor system and the force of the fixture at different inclination angles. Compare the experimental measurement data, under consideration of the mechanical properties of the system, making errors, and installation convenient prerequisites, the mechanical properties of the system are optimal when the thickness of the fixed plate is 30mm and the clamp tilt angle is 5 °.

scholarly journals Characterize traction–separation relation and interfacial imperfections by data-driven machine learning models

2021 ◽  
Vol 11 (1) ◽  
Author(s):  
Sanjida Ferdousi ◽  
Qiyi Chen ◽  
Mehrzad Soltani ◽  
Jiadeng Zhu ◽  
Pengfei Cao ◽  
...  
Keyword(s):  
Machine Learning ◽  
Mechanical Properties ◽  
Composite Materials ◽  
Structural Reliability ◽  
Mechanical Tests ◽  
Data Driven ◽  
Gradient Boosting ◽  
Material Interfaces ◽  
Element Analysis ◽  
Extreme Gradient Boosting

AbstractInterfacial mechanical properties are important in composite materials and their applications, including vehicle structures, soft robotics, and aerospace. Determination of traction–separation (T–S) relations at interfaces in composites can lead to evaluations of structural reliability, mechanical robustness, and failures criteria. Accurate measurements on T–S relations remain challenging, since the interface interaction generally happens at microscale. With the emergence of machine learning (ML), data-driven model becomes an efficient method to predict the interfacial behaviors of composite materials and establish their mechanical models. Here, we combine ML, finite element analysis (FEA), and empirical experiments to develop data-driven models that characterize interfacial mechanical properties precisely. Specifically, eXtreme Gradient Boosting (XGBoost) multi-output regressions and classifier models are harnessed to investigate T–S relations and identify the imperfection locations at interface, respectively. The ML models are trained by macroscale force–displacement curves, which can be obtained from FEA and standard mechanical tests. The results show accurate predictions of T–S relations (R2 = 0.988) and identification of imperfection locations with 81% accuracy. Our models are experimentally validated by 3D printed double cantilever beam specimens from different materials. Furthermore, we provide a code package containing trained ML models, allowing other researchers to establish T–S relations for different material interfaces.

scholarly journals Design of thermoformable three dimensional-printed PLA cast for fractured wrist

2022 ◽  
Vol 1217 (1) ◽  
pp. 012002
Author(s):  
N P Sorimpuk ◽  
W H Choong ◽  
B L Chua
Keyword(s):  
Mechanical Properties ◽  
3D Printing ◽  
Three Dimensional ◽  
Maximum Load ◽  
Production Method ◽  
Patient Specific ◽  
Element Analysis ◽  
Plaster Of Paris ◽  
3D Printed ◽  
Printing Method

Abstract Patient specific plastic cast for broken limbs has been developed recently in pharmaceutical field through three-dimensional (3D) printing method. However, the production of a 3D printed cast through normal 3D printing method is time consuming compared to conventional plaster casting. In this study, a design of ventilated structured thermoformable 3D-printed polylactic acid (PLA) cast was produced as an alternative for the 3D printed cast production method. This design was initially printed in a flat shape and then transformed into a cast which can be fitted to the user’s arm by using heat and external force. Finite Element Analysis (FEA) method was used to assess the mechanical properties of the proposed cast. In this analysis, thethermoformable design was exerted with a distributed force of 400 N, which is larger than the loading conditions encountered by human in their daily life. The mechanical properties of the thermoformable PLA cast such as local displacement under a specific load, maximum load, and stress were evaluated. Results were compared with the mechanical properties of Plaster of Paris cast. The results obtained from the FEA indicates that at the same layer thickness, the thermoformable 3D-printed PLA cast is stronger than the Plaster of Paris cast.

Sign in / Sign up

Export Citation Format

Share Document