Influence Diagrams Research Articles

Digital transformation and corporate strategy

The term digital transformation has been widely discussed and extensively promoted both to and by companies in every industry to become overused. While generally accepted as a boon and often marketed as the only alternative to extinction for a firm, digital transformation remains a significant challenge for those assigned to implement it. We offer an alternative narrative digital transformation, which we argue is better described as going «more digital». The extent to which going more digital becomes transformative depends on the industry and the ability of a firm to change the business model and organizational skills, and corporate culture. We extend the discussion on why digital transformation is in most cases unrelated to business strategy and provide examples of when and how it may become part of a corporate strategy toolbox. A simplified model of a business as a network of processes and decisions helps illustrate why digital technologies first spread in the process domain but can become strategic and transformative when adopted to the decision domain. The latter also brings yet unquantifiable risks and uncertainties. We suggest future research to focus on the value created by going «most digital» in both process and decision domains and risks to companies and the economy.

Path-Specific Objectives for Safer Agent Incentives

We present a general framework for training safe agents whose naive incentives are unsafe. As an example, manipulative or deceptive behaviour can improve rewards but should be avoided. Most approaches fail here: agents maximize expected return by any means necessary. We formally describe settings with `delicate' parts of the state which should not be used as a means to an end. We then train agents to maximize the causal effect of actions on the expected return which is not mediated by the delicate parts of state, using Causal Influence Diagram analysis. The resulting agents have no incentive to control the delicate state. We further show how our framework unifies and generalizes existing proposals.

EE13 Converting Discrete Event Simulation Networks (DESNETS) into Dice Models

Discrete event simulation (DES) is a formalism widely used for cost-effectiveness analysis (CEA). Discretely Integrated Condition Event (DICE) Simulation is a modelling framework for CEA. It is based on two concepts: conditions, which reflect aspects that persist, and events, which occur at a point in time. The implementation most used is an Excel add-on. DICE is widely accepted because it is transparent and simplifies the coding process. Discrete event simulation networks (DESnets) are a new type of probabilistic graphical model, somewhat similar to Bayesian networks and influence diagrams.

“Everything looks the same”: wayfinding behaviour and experiences of stroke inpatients in rehabilitation clinics

ABSTRACT Purpose To examine stroke inpatients’ real-life wayfinding behaviour and how the built environment of rehabilitation clinics might influence their behaviour and experiences. Methods Stroke inpatients in seven rehabilitation clinics were observed (n = 70), each over the course of 12 consecutive hours. Their paths through the clinic and the locations of encountered wayfinding-related events were mapped on the floor plans and described in the written notes. The observations were supplemented by a survey asking patients about their wayfinding experiences. Results For a third of observed patients, at least one wayfinding-related event was observed on the observation day, and 50% of patients reported getting lost in their clinic at least once. Most wayfinding-related events occurred between patient rooms and therapy rooms, and patients frequently relied on backtracking or the help of the staff to find their way. Clinics’ layout organization was found to play a role in the wayfinding behaviour of patients. Conclusions Wayfinding is a common challenge that stroke inpatients encounter in rehabilitation clinics. Avoiding multiple decision nodes on the paths between patient rooms and therapy rooms and creating distinct identities for corridor segments in the decision nodes and the areas in front of elevators would likely improve wayfinding performance.

Towards a machine learning-aided metaheuristic framework for a production/distribution system design problem

Recent advances have seen vast success in the application of metaheuristics in NP-hard combinatorial problems. A generic metaheuristic design usually consists of three core elements that jointly determine the algorithm performance, including an initial candidate solution, a guided search procedure, and a fitness function that approximates the objective value. This paper proposes a data-driven metaheuristic (DDMH) framework that leverages the predictive power of machine learning models, which exploit location information and mine structural knowledge of a supply chain network for intelligent decision making. Specifically, the proposed framework offers three performance boosters, including an initial solution heuristic, a narrowed search space, and an efficient learning-based fitness function. The framework can be readily integrated into existing MHs. As a case study, we apply DDMH to a production/distribution network design problem. Experimental results show that the DDMH outperforms the traditional MHs with better solution quality and comparable running time, especially for hard problems.

A Framework for Integrating Continuous Glucose Monitor-Derived Metrics into Economic Evaluations in Type 1 Diabetes.

Economic models in type 1 diabetes have relied on a change in haemoglobin A1c as the link between the blood glucose trajectory and long-term clinical outcomes, including microvascular and macrovascular disease. The landscape has changed in the past decade with the availability of regulatory approved, accurate and convenient continuous glucose monitoring devices and their ability to track patients' glucose levels over time. The data emerging from continuous glucose monitoring have enriched the clinical understanding of the disease and indirectly of patients' behaviour. This has triggered the development of new measures proposed to better define the quality of glycaemic control, beyond haemoglobin A1c. The objective of this paper is to review recent developments in clinical knowledge brought into focus with the application of continuous glucose monitoring devices, and to discuss potential approaches to incorporate the concepts into economic models in type 1 diabetes. Based on a targeted review and a series of multidisciplinary workshops, an influence diagram was developed that captures newer concepts (e.g. continuous glucose monitoring metrics) that can be integrated into economic models and illustrates their association with more established concepts. How the additional continuous glucose monitoring-based indicators of glycaemic control may contribute to economic modelling beyond haemoglobin A1c, and more accurately reflect the economic value of novel type 1 diabetes treatments, is discussed.

The Decisive Network: Magnum Photos and the Postwar Image Market

The Decisive Network: Magnum Photos and the Postwar Image Market

Using Social Networks to Sample Migrants and Study the Complexity of Contemporary Immigration: An Evaluation Study.

We test the effectiveness of a link-tracing sampling approach-network sampling with memory (NSM)-to recruit samples of rare immigrant populations with an application among Chinese immigrants in the Raleigh-Durham area of North Carolina. NSM uses the population network revealed by data from the survey to improve the efficiency of link-tracing sampling and has been shown to substantially reduce design effects in simulated sampling. Our goals are to (1) show that it is possible to recruit a probability sample of a locally rare immigrant group using NSM and achieve high response rates; (2) demonstrate the feasibility of the collection and benefits of new forms of network data that transcend kinship networks in existing surveys and can address unresolved questions about the role of social networks in migration decisions, the maintenance of transnationalism, and the process of social incorporation; and (3) test the accuracy of the NSM approach for recruiting immigrant samples by comparison with the American Community Survey. Our results indicate feasibility, high performance, cost-effectiveness, and accuracy of the NSM approach to sample immigrants for studies of local immigrant communities. This approach can also be extended to recruit multisite samples of immigrants at origin and destination.

Computerized history taking improves HEART score determination in acute chest pain patients

Abstract Funding Acknowledgements Type of funding sources: Foundation. Main funding source(s): Robert Bosch Stiftung (Stuttgart, Germany) Region Stockholm (ALF project; Stockholm, Sweden) Introduction Chest pain is a common chief complaint in emergency departments (EDs). The Acute Cardiovascular Care Association recommends the use of History, ECG, Age, Risk factors, and Troponin (HEART) score for risk stratification. In hectic and overcrowded EDs computerized history taking (CHT) is one structured way for collecting the components of HEART score, which include medical history (H- and R-variables). Purpose To determine the proportion of acute chest pain patients where CHT can be used to calculate HEART score, and what interrater reliability the included variables have with physician acquired medical history. Methods Prospective cohort study with acute chest pain patients self-reporting medical histories using a CHT program (Clinical Expert Operating System, CLEOS) on a tablet. CLEOS, with >17,000 decision nodes, mimics a physician interview and tailors each interview continuously depending on previous answers. Clinically stable women and men aged >18 years with a chief complaint of chest pain and non-diagnostic ECG or serum markers for acute coronary syndrome were enrolled. Patients unable to carry out a CHT interview (e.g., severe impaired vision or confusion) were excluded. As recommended in regional guidelines a modified HEART score using the traditional classification of anginal symptoms, i.e. 1) central chest pain, 2) provoked by physical exertion and/or emotional stress, and 3) relieved by rest and/or nitrates, and traditional risk factors was used (see Table 1). Observations to each discrete H- and R-variable and medical history data were extracted from electronic health records (EHR) and the CHT database by the research staff. Cohen’s kappa statistics and interpretation according to Landis & Koch was used to describe the interrater reliability. Results A total of 1,000 consecutive patients were enrolled in one tertiary university hospital ED during 2017–2019 (mean age 55±17 years; 54% women), comparable to the general chest pain population (mean age 58±19 years; 50% women). A complete HEART score could be calculated in 60% of patients (H-variable 91%; R-variable 60%). Registered observations from EHR and CHT and interrater reliability data are presented in Table 1. Interrater reliability for the H-variable (classical anginal symptoms) was slight to moderate (kappa 0.19-0.44) and the R-variable (traditional risk factors) was moderate to almost perfect (kappa 0.56-0.88). Conclusions HEART score could be calculated in a majority of acute chest pain patients by the use of CHT. Interrater reliability was high for traditional risk factors but low to moderate for classical anginal symptoms. Our ongoing studies assess whether CHT combined with a risk score such as HEART score in an acute setting can be useful for improved risk stratification and, more important, to improve clinical outcomes.

A probabilistic framework for risk management and emergency decision-making of marine oil spill accidents

Offshore oil spills may pose a severe threat to marine ecological environment. In this paper, a new methodology based on Bayesian network (BN) and Influence Diagram (ID) is developed for risk management and emergency decision-making of marine oil spill accidents. The methodology integrates risk management before accident and emergency response after accident, which can balance risk and cost, and render an optimal decision-making. Marine oil spill scenarios including root causations, intermediate and consequent events are identified and modeled using BN considering the dependencies and multi-state of incident process nodes. The probabilities of offshore oil spill incident and the resulting ecological disasters are estimated. The prevention and mitigation measures marine oil spill incidents are identified and added to BN for developing Bayesian ID model, in which the cost and utility of implementing each safety measure are considered. Bayesian ID model can estimate the cost and utility of all safety strategies, and the optimal risk management and emergency strategy are determined by balancing the cost and utility. A case study of marine oil spill accident due to subsea oil pipeline leak is used to illustrate the methodology. It is observed that the methodology can efficiently support the decision-making of oil and gas sector in risk management and emergency response of offshore oil spill accidents.

Dynamic path learning in decision trees using contextual bandits

We present a novel online decision-making solution, where the optimal path of a given decision tree is dynamically found based on the contextual bandits analysis. At each round, the learner finds a path in the decision tree by making a sequence of decisions following the tree structure and receives an outcome when a terminal node is reached. At each decision node, the environment information is observed to hint on which child node to visit, resulting in a better outcome. The objective is to learn the context-specific optimal decision for each decision node to maximize the accumulated outcome. In this paper, we propose Dynamic Path Identifier (DPI), a learning algorithm where the contextual bandit is applied to every decision node, and the observed outcome is used as the reward of the previous decisions of the same round. The technical difficulty of DPI is the high exploration challenge caused by the width (i.e., the number of paths) of the tree as well as the large context space. We mathematically prove that DPI’s regret per round approached zero as the number of the rounds approaches infinity. We also prove that the regret is not a function of the number of paths in the tree. Numerical evaluations are provided to complement the theoretical analysis.

Ensuring representativity of scenario sets: The importance of exploring unknown unknowns

This paper presents methodological results from a scenario project about the linkages between changing geopolitics and sustainability. If one wishes to take wise long-term investment decisions in such domains one should strive for turning important ‘unknown unknowns’ into analysable ‘known unknowns’. As outlined in the paper we see this capability as one of several requirements for representative scenario sets in the mentioned type of planning situation. Based on these requirements, a novel scenario methodology called structural variables intuitive logics (SV-IL) was developed. Starting from an in-depth critique of the well-known ‘scenario matrix’ technique, SV-IL brings together the intuitive logics approach for combining imagination and judgment, with elements of cross impact methodology to further enhance representativity. The core element is a workshop format where small expert teams are tasked to make sense of well-defined scenario seeds by iteratively developing event-based storylines and their resulting end-states. This allows path dependence to emerge based on each scenario’s storyline. The states are modelled by influence diagrams with circular causality (feedback) a key element that helps generate persistent scenarios forming useful starting points for understanding potential future societies. Making the well-defined seeds diverse led to distinctive scenarios which benefits broad exploration of potential futures.

Decision analysis on disposal of large quantities of excavated soft soil in abandoned mines using a Bayesian network

ABSTRACT Disposal of the excess excavated soft soils produced in construction is a contemporary engineering challenge. A win-win solution could be found through utilising much excavated soil during reclamation of abandoned mines, and an evaluation model is necessary to realise this solution. This study aimed to propose a decision-making framework for large-scale disposal of excavated soft soil during mine reclamation using a Bayesian network (BN). Utilising multi-facet information, a decision-making path considering engineering and economic factors, and engineering risks was established. Thereafter, the practicability of backfilling much excavated soft soil into 215 abandoned opencast mines located in Wenzhou, China was evaluated. Based on the decision-making path and information of the 215 mines, a BN including 22 nodes was established with the practicability of large-scale soil disposal in mines as the decision node. Consequently, 40 valid mines were evaluated to be able to absorb 5 × 107 m3 soil, around 50% of the total excavated soil from the municipal districts of Wenzhou during 2020–2024. The BN evaluation showed high reliability with accuracy of all the nodes >70% and that of the decision node >90%. Five input nodes were identified to be high-impact for evaluation of the practicability and total fill volume.

IFC BIM Model Enrichment with Space Function Information Using Graph Neural Networks

The definition of room functions in Building Information Modeling (BIM) using IfcSpace entities is an important quality requirement that is often not fulfilled. This paper presents a three-step method for enriching open BIM representations based on Industry Foundation Classes (IFC) with room function information (e.g., kitchen, living room, foyer). In the first step, the geometric algorithm for detecting and defining IfcSpace entities and injecting them into IFC models is presented. After deriving the IfcSpaces, a geometric method for calculating the graph of connections between spaces based on accessibility is described; this information is not explicitly stored in IFC models. In the final step, a graph convolution-based neural network using the accessibility graph to classify the IfcSpace entities is described. Local node features are automatically extracted from the geometry and neighboring elements. With the help of a Graph Convolutional Network (GCN), the connection and spatial context information is utilized by the neural network for the classification decision, in addition to the local features of the spaces which are more commonly used. To evaluate the classification accuracy, the model was tested on a set of residential building IFC models. A weighted version of the common GCN was implemented and tested, resulting in a slight improvement in the classification accuracy.

Unsupervised machine learning to classify crystal structures according to their structural distortion: A case study on Li-argyrodite solid-state electrolytes

High-throughput approaches in computational materials discovery often yield a combinatorial explosion that makes the exhaustive rendering of complete structural and chemical spaces impractical. A common bottleneck when screening new compounds with archetypal crystal structures is the lack of fast and reliable decision-making schemes to quantitatively classify the computed candidates as inliers or outliers (too distorted structures). Machine learning-aided workflows can solve this problem and make geometrical optimization procedures more efficient. However, for this to occur, there is still a lack of appropriate combinations of suitable geometrical descriptors and accurate unsupervised models which are capable of accurately differentiating between systems with subtle structural changes. Here, considering as a case study the compositional screening of cubic Li-argyrodites solid electrolytes, we tackle this problem head on. We find that Steinhardt order parameters are very accurate descriptors of the cubic argyrodite structure to train a range of common unsupervised outlier detection models. And, most importantly, the approach enables us to automatically classify crystal structures with uncertainty control. The resulting models can then be used to screen computed structures with respect to an user-defined error threshold and discard too distorted structures during geometrical optimization procedures. Implemented as a decision node in computer-aided materials discovery workflows, this approach can be employed to perform autonomous high-throughput screening methods and make the use of computational and data storage resources more efficient.

Black-White incentive inequality for college persistence

Despite similar educational aspirations, black students persist in higher education at much lower rates than white undergraduates. This paper advances a theoretical explanation for the racial gap in persistence by examining whether the differential attrition in college reflects contrasting incentives for educational persistence. To account for the highly unequal hurdles faced by black men and women in college and in the labor market, we propose a method that addresses race-gender-specific opportunity structures in both institutions simultaneously. This approach is based on forward-looking estimates of outcomes where students draw information from their race-gender reference group ahead of them. The model estimates the earnings payoffs of persistence separately for each race-gender group at three consecutive educational decision nodes: at high school graduation, college entry, and after one year in college. We subsequently apply one version of this model to data from the American Community Surveys (2001–2017), calculating the absolute and relative incentives for educational persistence across racial groups. In addition to large dollar earnings differentials, the analyses reveal striking racial gaps of the relative incentives to stay enrolled: “incentive inequality.” This incentive race gap is largest at the earliest stages of the higher education career—high school graduation and college entry—where the black undergraduate dropout rate is highest. Our findings have substantive and methodological implications for situations where returns to investments are unequal across groups affected by discrimination.

Therapeutic drug monitoring in inflammatory bowel disease: At the right time in the right place.

Therapeutic drug monitoring (TDM) was one of most sought-after objective tools to determine therapeutic efficiency of different biologics and its role in the management of patients with inflammatory bowel disease (IBD) was regarded with great anticipation. But implementation of the TDM in clinical practice was challenged by several factors including uncertainty of the optimal cut-off values, assay variable sensitivity in detecting drug levels and antibodies and, most importantly, individual pharmacokinetics. While reactive TDM was embraced in clinical practice as a useful tool in assessing lack of response to therapy, the utility of proactive TDM in managing IBD therapy is still challenged by the lack of consistency between evidence. Described here, there are four groups of IBD patients for whom proactive TDM has the potential to greatly impact their therapeutic outcomes: Patients with perianal Crohn’s disease, patients with severe ulcerative colitis, pregnant women with IBD and children. As the future of IBD management moves towards personalizing treatment, TDM will be an important decision node in a machine learning based algorithm predicting the best strategy to maximize treatment results while minimizing the loss of response to therapy.

A cumulative risk and sustainability index for pavements

Pavement sustainability is measured using various methodologies which often focus on only certain aspects of sustainability. Methods to quantify and unify holistic sustainability in terms of economic, environmental and social consequences are lacking. In previous papers, the major findings of the contribution of construction, maintenance and use to pavement sustainability were presented. The objective of this paper is to present a method of quantification using a risk and sustainability index that can compare the design and maintenance options of pavement systems. The model utilises Markov chain Monte Carlo simulations applied to a conceptualised influence diagram to determine the probability of sustainability using a cumulative approach. Previously reported results are provided which form the basis of the model, detailing economic, environmental and social results of pavement alternatives, followed by the development of a probability model to quantify holistic pavement sustainability. In conclusion, even though sustainability is complex, simplification through simulation is achievable.

Integrated approaches to testing and assessment for grouping nanomaterials following dermal exposure

Exposure to different nanoforms (NFs) via the dermal route is expected in occupational and consumer settings and thus it is important to assess their dermal toxicity and the contribution of dermal exposure to systemic bioavailability. We have formulated four grouping hypotheses for dermal toxicity endpoints which allow NFs to be grouped to streamline and facilitate risk assessment. The grouping hypotheses are developed based on insight into how physicochemical properties of NFs (i.e. composition, dissolution kinetics, size, and flexibility) influence their fate and hazard following dermal exposure. Each hypothesis is accompanied by a tailored Integrated Approach to Testing and Assessment (IATA) that is structured as a decision tree and tiered testing strategies (TTS) for each relevant question (at decision nodes) that indicate what information is needed to guide the user to accept or reject the grouping hypothesis. To develop these hypotheses and IATAs, we gathered and analyzed existing information on skin irritation, skin sensitization, and dermal penetration of NFs from the published literature and performed experimental work to generate data on NF dissolution in sweat simulant fluids. We investigated the dissolution of zinc oxide and silicon dioxide NFs in different artificial sweat fluids, demonstrating the importance of using physiologically relevant conditions for dermal exposure. All existing and generated data informed the formulation of the grouping hypotheses, the IATAs, and the design of the TTS. It is expected that the presented IATAs will accelerate the NF risk assessment for dermal toxicity via the application of read-across.

Assessment of uncertain alternatives for co-located aquaculture and offshore wind farm in tasmania

This study investigated a set of 36 alternatives of co-located aquaculture and an offshore wind farm in three potential sites across Tasmania. We assessed the alternatives against technological aspects, environmental aspects, and site-selection factors through the development of a Bayesian network and an influence diagram such that the uncertainties of each possible vector of consequences were effectively assessed for each alternative. We structured the consequence space as a three-level hierarchical tree and constructed the multi-attribute utility function under various independence conditions. We described in detail the assessment of the scaling constants in the multi-attribute utility function. The analysed alternatives were ranked based on the maximum expected utility rule to outline the best one. We performed one-dimensional and multi-dimensional sensitivity analyses, to test how the solution is impacted by the subjectively assessed five-dimensional utility function and showed that the constructed function is rather robust within a very wide range of values. Finally, we outlined some future aspects of development and implementation for the Bayesian-network-influence diagram based multi-attribute analysis as a decision-support system tool for offshore Blue Growth projects with different types of production systems.