Hybrid Bayesian Network Models of Spinal Injury and Slip/Fall Events

Richard Hughes

doi:10.3390/app10144834

Hybrid Bayesian Network Models of Spinal Injury and Slip/Fall Events

Applied Sciences ◽

10.3390/app10144834 ◽

2020 ◽

Vol 10 (14) ◽

pp. 4834

Author(s):

Richard Hughes

Keyword(s):

Bayesian Networks ◽

Spinal Injury ◽

Network Models ◽

Probabilistic Methods ◽

Deterministic Analysis ◽

Legal Proceedings ◽

Standard Of Proof ◽

Biomechanical Models ◽

Expert Opinions ◽

Engineering Mechanics

Background: Biomechanists are often asked to provide expert opinions in legal proceedings, especially personal injury cases. This often involves using deterministic analysis methods, although the expert is expected to opine using a civil standard of “more likely than not” that is inherently probabilistic. Methods: A method is proposed for converting a class of deterministic biomechanical models into hybrid Bayesian networks that produce a probability well suited for addressing the civil standard of proof. The method was developed for spinal injury during lifting. Its generalizability was assessed by applying it to slip and fall events based on the coefficients of friction at the shoe–floor interface. Results: The proposed method is shown to be generalizable beyond lifting by applying it to a slip and fall event. Both the lifting and slip and fall models showed that incorporating evidence of injury could change the probabilities of critical quantities exceeding a threshold from “less likely than not” to “more likely than not.” Conclusions: The present work shows that it is possible to develop Bayesian networks for legal use based on laws of engineering mechanics and probabilistic descriptions of measurement error and human variability.

Download Full-text

STANDARDS OF PROOF: AID OR PITFALL?

Obiter ◽

10.17159/obiter.v31i3.12334 ◽

2021 ◽

Vol 31 (3) ◽

Author(s):

Andra le Roux-Kemp

Keyword(s):

General Public ◽

Natural Sciences ◽

Point Of View ◽

Pivotal Role ◽

Legal Proceedings ◽

Standard Of Proof ◽

Legal Science ◽

Standards Of Proof ◽

Responsible Party

During the course of legal proceedings, evidentiary material is analyzed and evaluated in order to make a final judgement whether the responsible party has discharged the onus of proof. The existence of a standard of proof against which the presiding officer can measure the evidence submitted consequently plays a pivotal role. This standard of proof (bewysmaatstaf) represents the standard of guilt in legal science and has also been described as a standard of conviction. The standard of proof does not pertain to the inherent qualities of evidentiary material, but rather to the degrees of conviction of the presiding officer in a particular case. The function of thestandard of proof is furthermore to provide presiding officers with a guideline/yardstick to measure the degree of conviction that the general public believe the presiding officer should have over the correctness of all the factual conclusions in the particular proceedings. In this article, the standard of proof in law will be discussed from a comparative point of view; different standards of proof from different jurisdictions will be considered and juxtaposed against similar standards used in the natural sciences.

Download Full-text

AIS-BN: An Adaptive Importance Sampling Algorithm for Evidential Reasoning in Large Bayesian Networks

Journal of Artificial Intelligence Research ◽

10.1613/jair.764 ◽

2000 ◽

Vol 13 ◽

pp. 155-188 ◽

Cited By ~ 93

Author(s):

J. Cheng ◽

M. J. Druzdzel

Keyword(s):

Bayesian Networks ◽

Bayesian Network ◽

Importance Sampling ◽

Network Models ◽

Evidential Reasoning ◽

The Other ◽

Sampling Algorithm ◽

Importance Function ◽

Sampling Algorithms ◽

Bayesian Network Models

Stochastic sampling algorithms, while an attractive alternative to exact algorithms in very large Bayesian network models, have been observed to perform poorly in evidential reasoning with extremely unlikely evidence. To address this problem, we propose an adaptive importance sampling algorithm, AIS-BN, that shows promising convergence rates even under extreme conditions and seems to outperform the existing sampling algorithms consistently. Three sources of this performance improvement are (1) two heuristics for initialization of the importance function that are based on the theoretical properties of importance sampling in finite-dimensional integrals and the structural advantages of Bayesian networks, (2) a smooth learning method for the importance function, and (3) a dynamic weighting function for combining samples from different stages of the algorithm. We tested the performance of the AIS-BN algorithm along with two state of the art general purpose sampling algorithms, likelihood weighting (Fung & Chang, 1989; Shachter & Peot, 1989) and self-importance sampling (Shachter & Peot, 1989). We used in our tests three large real Bayesian network models available to the scientific community: the CPCS network (Pradhan et al., 1994), the PathFinder network (Heckerman, Horvitz, & Nathwani, 1990), and the ANDES network (Conati, Gertner, VanLehn, & Druzdzel, 1997), with evidence as unlikely as 10^-41. While the AIS-BN algorithm always performed better than the other two algorithms, in the majority of the test cases it achieved orders of magnitude improvement in precision of the results. Improvement in speed given a desired precision is even more dramatic, although we are unable to report numerical results here, as the other algorithms almost never achieved the precision reached even by the first few iterations of the AIS-BN algorithm.

Download Full-text

Advanced Sequence Analysis

10.1093/oso/9780192897879.003.0011 ◽

2021 ◽

pp. 307-344

Author(s):

Magy Seif El-Nasr ◽

Truong Huy Nguyen Dinh ◽

Alessandro Canossa ◽

Anders Drachen

Keyword(s):

Sequence Analysis ◽

Bayesian Networks ◽

Recurrent Neural Networks ◽

Short Term Memory ◽

Markov Models ◽

Dynamic Bayesian Networks ◽

Probabilistic Methods ◽

Markov Logic ◽

Markov Decision ◽

Long Short Term Memory

This chapter discusses more advanced methods for sequence analysis. These include: probabilistic methods using classical planning, Bayesian Networks (BN), Dynamic Bayesian Networks (DBNs), Hidden Markov Models (HMMs), Markov Logic Networks (MLNs), Markov Decision Process (MDP), and Recurrent Neural Networks (RNNs), specifically concentrating on LSTM (Long Short-Term Memory). These techniques are all great but, at this time, are mostly used in academia and less in the industry. Thus, the chapter takes a more academic approach, showing the work and its application to games when possible. The techniques are important as they cultivate future directions of how you can think about modeling, predicting players’ strategies, actions, and churn. We believe these methods can be leveraged in the future as the field advances and will have an impact in the industry. Please note that this chapter was developed in collaboration with several PhD students at Northeastern University, specifically Nathan Partlan, Madkour Abdelrahman Amr, and Sabbir Ahmad, who contributed greatly to this chapter and the case studies discussed.

Download Full-text

Retrieving Medical Records Using Bayesian Networks

Medical Informatics ◽

10.4018/978-1-60566-050-9.ch172 ◽

2011 ◽

pp. 2274-2280

Author(s):

Luis M. De Campos

Keyword(s):

Information Retrieval ◽

Bayesian Networks ◽

Graphical Models ◽

Medical Records ◽

Network Models ◽

Boolean Expression ◽

Wide Range ◽

Structured Information Retrieval ◽

Structured Information ◽

Bayesian Network Models

Bayesian networks (Jensen, 2001) are powerful tools for dealing with uncertainty. They have been successfully applied in a wide range of domains where this property is an important feature, as in the case of information retrieval (IR) (Turtle & Croft, 1991). This field (Baeza-Yates & Ribeiro- Neto, 1999) is concerned with the representation, storage, organization, and accessing of information items (the textual representation of any kind of object). Uncertainty is also present in this field, and, consequently, several approaches based on these probabilistic graphical models have been designed in an attempt to represent documents and their contents (expressed by means of indexed terms), and the relationships between them, so as to retrieve as many relevant documents as possible, given a query submitted by a user. Classic IR has evolved from flat documents (i.e., texts that do not have any kind of structure relating their contents) with all the indexing terms directly assigned to the document itself toward structured information retrieval (SIR) (Chiaramella, 2001), where the structure or the hierarchy of contents of a document is taken into account. For instance, a book can be divided into chapters, each chapter into sections, each section into paragraphs, and so on. Terms could be assigned to any of the parts where they occur. New standards, such as SGML or XML, have been developed to represent this type of document. Bayesian network models also have been extended to deal with this new kind of document. In this article, a structured information retrieval application in the domain of a pathological anatomy service is presented. All the medical records that this service stores are represented in XML, and our contribution involves retrieving records that are relevant for a given query that could be formulated by a Boolean expression on some fields, as well as using a text-free query on other different fields. The search engine that answers this second type of query is based on Bayesian networks.

Download Full-text

SPINET: A Parallel Computing Approach to Spine Simulations

Scientific Programming ◽

10.1155/1996/607139 ◽

1996 ◽

Vol 5 (1) ◽

pp. 15-24 ◽

Cited By ~ 1

Author(s):

Peter G. Kropf ◽

Edgar F. A. Lederer ◽

Thomas Steffen ◽

Karl Guggisberg ◽

Jean-Guy Schneider ◽

...

Keyword(s):

Finite Element ◽

Parallel Computing ◽

Functional Programming ◽

High Performance ◽

Dynamic Models ◽

Spinal Injury ◽

Direct Method ◽

Finite Element Program ◽

Biomechanical Models ◽

Interdisciplinary Approaches

Research in scientitic programming enables us to realize more and more complex applications, and on the other hand, application-driven demands on computing methods and power are continuously growing. Therefore, interdisciplinary approaches become more widely used. The interdisciplinary SPINET project presented in this article applies modern scientific computing tools to biomechanical simulations: parallel computing and symbolic and modern functional programming. The target application is the human spine. Simulations of the spine help us to investigate and better understand the mechanisms of back pain and spinal injury. Two approaches have been used: the first uses the finite element method for high-performance simulations of static biomechanical models, and the second generates a simulation developmenttool for experimenting with different dynamic models. A finite element program for static analysis has been parallelized for the MUSIC machine. To solve the sparse system of linear equations, a conjugate gradient solver (iterative method) and a frontal solver (direct method) have been implemented. The preprocessor required for the frontal solver is written in the modern functional programming language SML, the solver itself in C, thus exploiting the characteristic advantages of both functional and imperative programming. The speedup analysis of both solvers show very satisfactory results for this irregular problem. A mixed symbolic-numeric environment for rigid body system simulations is presented. It automatically generates C code from a problem specification expressed by the Lagrange formalism using Maple.

Download Full-text

A Risk-Informed Approach to Leak-Before-Break Assessment of Pressure Tubes in CANDU Reactors

Journal of Pressure Vessel Technology ◽

10.1115/1.3148087 ◽

2010 ◽

Vol 132 (2) ◽

Cited By ~ 1

Author(s):

M. D. Pandey ◽

A. K. Sahoo

Keyword(s):

Computation Time ◽

Service Evaluation ◽

Probabilistic Methods ◽

Pressure Tube ◽

Deterministic Method ◽

Deterministic Analysis ◽

Candu Reactors ◽

Technical Requirements ◽

Pressure Tubes ◽

Leak Before Break

The leak-before-break (LBB) assessment of pressure tubes is intended to demonstrate that in the event of through-wall cracking of the tube, there will be sufficient time followed by the leak detection, for a controlled shutdown of the reactor prior to the rupture of the pressure tube. CSA Standard N285.8 (2005, “Technical Requirements for In-Service Evaluation of Zirconium Alloy Pressure Tubes in CANDU Reactors,” Canadian Standards Association) has specified deterministic and probabilistic methods for LBB assessment. Although the deterministic method is simple, the associated degree of conservatism is not quantified and it does not provide a risk-informed basis for the fitness for service assessment. On the other hand, full probabilistic methods based on simulations require excessive amount of information and computation time, making them impractical for routine LBB assessment work. This paper presents an innovative, semiprobabilistic method that bridges the gap between a simple deterministic analysis and complex simulations. In the proposed method, a deterministic criterion of CSA Standard N285.8 is calibrated to specified target probabilities of pressure tube rupture based on the concept of partial factors. This paper also highlights the conservatism associated with the current CSA Standard. The main advantage of the proposed approach is that it retains the simplicity of the deterministic method, yet it provides a practical, risk-informed basis for LBB assessment.

Download Full-text

Bayesian Network: Probabilistic Reasoning, Statistical Learning, and Applications

Journal of Advanced Computational Intelligence and Intelligent Informatics ◽

10.20965/jaciii.2004.p0093 ◽

2004 ◽

Vol 8 (2) ◽

pp. 93-99 ◽

Cited By ~ 1

Author(s):

Yoichi Motomura ◽

Keyword(s):

Information Processing ◽

Bayesian Networks ◽

Bayesian Network ◽

Statistical Learning ◽

Probabilistic Models ◽

Probabilistic Reasoning ◽

Network Models ◽

Intelligent Information Processing ◽

Intelligent Information ◽

Bayesian Network Models

Bayesian networks are probabilistic models that can be used for prediction and decision-making in the presence of uncertainty. For intelligent information processing, probabilistic reasoning based on Bayesian networks can be used to cope with uncertainty in real-world domains. In order to apply this, we need appropriate models and statistical learning methods to obtain models. We start by reviewing Bayesian network models, probabilistic reasoning, statistical learning, and related researches. Then, we introduce applications for intelligent information processing using Bayesian networks.

Download Full-text

Fraud Detection of Bulk Cargo Theft in Port Using Bayesian Network Models

Applied Sciences ◽

10.3390/app10031056 ◽

2020 ◽

Vol 10 (3) ◽

pp. 1056

Author(s):

Rongjia Song ◽

Lei Huang ◽

Weiping Cui ◽

María Óskarsdóttir ◽

Jan Vanthienen

Keyword(s):

Bayesian Networks ◽

Bayesian Network ◽

Predictive Models ◽

Domain Knowledge ◽

Fraud Detection ◽

Network Models ◽

Risk Elements ◽

Data Driven Approach ◽

Binary Classifiers ◽

Cargo Theft

The fraud detection of cargo theft has been a serious issue in ports for a long time. Traditional research in detecting theft risk is expert- and survey-based, which is not optimal for proactive prediction. As we move into a pervasive and ubiquitous paradigm, the implications of external environment and system behavior are continuously captured as multi-source data. Therefore, we propose a novel data-driven approach for formulating predictive models for detecting bulk cargo theft in ports. More specifically, we apply various feature-ranking methods and classification algorithms for selecting an effective feature set of relevant risk elements. Then, implicit Bayesian networks are derived with the features to graphically present the relationship with the risk elements of fraud. Thus, various binary classifiers are compared to derive a suitable predictive model, and Bayesian network performs best overall. The resulting Bayesian networks are then comparatively analyzed based on the outcomes of model validation and testing, as well as essential domain knowledge. The experimental results show that predictive models are effective, with both accuracy and recall values greater than 0.8. These predictive models are not only useful for understanding the dependency between relevant risk elements, but also for supporting the strategy optimization of risk management.

Download Full-text