Risk Modeling Research Articles

Achieving successful gas plant construction: A risk modelling framework with quantitative risk-based approach to schedule management

The construction of gas plants often experiences delays caused by various factors, which can lead to significant financial and operational losses. This research aims to develop an accurate risk model to improve the schedule performance of gas plant projects. The model uses Quantitative Risk Analysis (QRA) and Monte Carlo simulation methods to identify and measure the risks that most significantly impact project schedule performance. A comprehensive literature review was conducted to identify the risk variables that may cause delays. The risk model, pre-simulation modeling, result analysis, and expert validation were all developed using a Focused Group Discussion (FGD). Primavera Risk Analysis (PRA) software was used to perform Monte Carlo simulations. The simulation output provides information on probability distribution, histograms, descriptive statistics, sensitivity analysis, and graphical results that aid in better understanding and decision-making regarding project risks. The research results show that the simulated project completion timeline after mitigation suggested an acceleration of 61–65 days compared to the findings of the baseline simulation. This demonstrates that activity-based mitigation has a major influence on improving schedule performance. This research makes a significant contribution to addressing project delay issues by introducing an innovative and effective risk model. The model empowers project teams to proactively identify, measure, and mitigate risks, thereby improving project schedule performance and delivering more successful projects.

Development and validation of a clinical risk model for postoperative outcome in newly diagnosed glioblastoma: a report of the RANO resect group.

Following surgery, patients with newly diagnosed glioblastoma frequently enter clinical trials. Nuanced risk assessment is warranted to reduce imbalances between study arms. Here, we aimed (I) to analyze the interactive effects of residual tumor with clinical and molecular factors on outcome and (II) to define a postoperative risk assessment tool. The RANO resect group retrospectively compiled an international, seven-center training cohort of patients with newly diagnosed glioblastoma. The combined associations of residual tumor with molecular or clinical factors and survival were analyzed, and recursive partitioning analysis was performed for risk modeling. The resulting model was prognostically verified in a separate external validation cohort. Our training cohort compromised 1003 patients with newly diagnosed isocitrate dehydrogenase-wildtype glioblastoma. Residual tumor, O6-methylguanine DNA methyltransferase (MGMT) promotor methylation status, age, and postoperative KPS were prognostic for survival and incorporated into regression tree analysis. By individually weighting the prognostic factors, an additive score (range, 0-9 points) integrating these four variables distinguished patients with low (0-2 points), intermediate (3-5 points), and high risk (6-9 points) for inferior survival. The prognostic value of our risk model was retained in treatment-based subgroups and confirmed in an external validation cohort of 258 patients with glioblastoma. Compared to previously postulated models, goodness-of-fit measurements were superior for our model. The novel RANO risk model serves as an easy-to-use, yet highly prognostic tool for postoperative patient stratification prior to further therapy. The model may serve to guide patient management and reduce imbalances between study arms in prospective trials.

Modeling of Soil Erosion Risk Utilizing Morphometric Analysis and Geospatial Approach in Lake Nakuru, Kenya

Modeling of Soil Erosion Risk Utilizing Morphometric Analysis and Geospatial Approach in Lake Nakuru, Kenya

Patient specific numerical hemodynamics for postoperative risk assessment: series case study of EC-IC cerebral bypass

Abstract The study is devoted to the hemodynamics during cerebral vascular bypass surgery for the treatment of cerebral aneurysms in two patients. The location, morphological characteristics and treatment approaches of the patients were similar, but different outcomes were observed as a result of the performed microsurgical procedures . Computational approach was used to analyze the hemodynamic differences of aneurysms, treated via extra-intra cranial (EC-IC) cerebral bypass shunt. The paper presents a new criterion based on the energy parameters of healthy compartment of cerebral circulation. The applied approach demonstrates a new effective method of preoperative risk modelling for medical decision-making.

Developing Predictive Models for Financial Stability: Integrating Behavioral Analytics into Credit Risk Management

In this article, its author takes an attempt at investigating the use of behavioral analytics to complement credit risk management in the world of finance. Current mainstream credit risk models very much depend on conventional methods that incorporate historical financial information like credit scores and income while ignoring client behavioral contemporaneous patterns essential for modeling their risk levels. By this, the transactional patterns; payment behaviour and digital activities included besides the traditional financial parameters allows the financial institution to make superior models that gives better results on credit worthiness and risk identification earlier. It is crucial for both better risk assessment and for embracing more inclusion within lending networks and financial creativity. Analyzing the informational component of behavioral data as well as its applicability in credit risk modeling and considering the difficulties of implementing it in practice, the article provides an outline of the approaches to telemetry integration.

Artificial Intelligence in public health: A case study

Artificial Intelligence technology has become an innovative tool in global healthcare that helps to manage various problems of scale and complexity. Its use in public health is multifaceted and multivariate extending across early outbreak detection, risk modeling and prediction, efficient vaccine logistics and administration, reducing health disparities, aiding assist in diagnosis, everyday monitoring of chronic diseases, telemedicine and virtual chatbots especially for elderly care, mental health and oncology patients, and community engagement. Despite these huge generational advantages, multiple challenges exist in data quality, security and misuse, individual privacy, contextual application, accountability and oversight and remain primary concerns for universal adoption of artificial intelligence. There is a need for safeguards and a global regulatory framework to responsibly harvest the benefits of artificial intelligence for the advancement of public and community health.

Implementation of polygenic risk scores in secondary prevention of (breast) cancer

Abstract Utilizing polygenic risk scores to inform secondary prevention of breast cancer via population screening programs is one of the pioneering applications of genomics in public health applications. Over the last decade, polygenic risk scores for breast cancer have been developed and validated in various European populations (AUC ± 0.63, OR per Z unit ± 1.6, depending on the population). In some countries, the polygenic risk score has been integrated with decision modeling tools, such as CanRisk in the Netherlands, and low-level implementation into health and care systems has started. Implementation strategies vary, but generally consider initial implementation into high-risk populations, such as breast cancer families, in specialized clinical setting, such as academic medical centres, or specialized cancer clinics, referring relatives of patients to population screening based on comprehensive risk modelling, including polygenic risk scores. This route may be followed by other disease fields as well. The presentations outlines the current evidence and considerations on breast cancer genetic risk modelling, clinical pilot studies and the outline of step-wise implementation of such models to benefit secondary prevention programs. We start from the perspective of a single hospital, to the needed steps to upscale to national populations, both in diversity of these populations, as the throughput of genetic testing and counselling, until the harmonization and recalibration of various European healthcare models. We draw from examples of the Genotyping on all patients (GOALL), building cancer health platforms (CanHeal) and Genome of Europe (GoE) consortia.

CMR phenotypes and clinical outcomes in NYHA class I versus II-IV hypertrophic cardiomyopathy: identifying high-risk asymptomatic patients

Abstract Background Cardiac myosin inhibition therapy can deliver symptomatic benefit for NYHA class≥II patients with obstructive hypertrophic cardiomyopathy (HCM) and is being actively explored for non-obstructive phenotypes. However, expanding interest in clinical outcome reduction supports growing need to identify high risk cohorts across both symptomatic and asymptomatic cohorts. We aimed to examine phenotypic characteristics of NYHA-I versus NYHA≥II patients with HCM and their association with future outcomes. Methods 727 patients with HCM from the CIROC Registry who completed simultaneous CMR and patient health assessments, inclusive of NYHA evaluation and quality of life surveys, were studied. Baseline clinical and CMR characteristics were compared between NYHA-I and NYHA≥II cohorts, followed by risk modelling for the prediction of major adverse cardiovascular events (MACE), defined as: all-cause mortality, heart failure hospitalization, ventricular arrhythmia, new onset atrial fibrillation, or stroke. Cox models were constructed to identify independent predictors of MACE. Results Of the 727 patients, 546 (75%) reported NYHA-I and 181 (25%) NYHA≥II symptoms. Baseline characteristics are shown in Table 1. Compared to NYHA≥II, NYHA-I patients were younger, more likely male, and had a lower prevalence of diabetes and hypertension. No meaningful differences in CMR chamber volumes, function, left ventricular (LV) mass, or fibrosis burden were observed. Over a median follow up of 3.7 years, 109 patients (15%) experienced MACE: 12% in NYHA-I and 23% in NYHA≥II sub-groups (p<0.001). Baseline LV mass z-scores (determined from sex-matched, BSA-indexed SCMR healthy reference values) were independently associated with MACE by multivariable Cox modelling. An optimal LV mass z-score threshold of 3.9 for prediction of MACE was identified and combined with NYHA status to stratify patients into 4 categories. Kaplan-Meier curves showed patients with LV Mass z-scores >3.9 experienced worse event-free survival in both NYHA sub-groups (Figure 1a). Cox modelling in the overall cohort (Figure 1b) showed NYHA-I patients with LV Mass z-scores >3.9 experienced a 2.2-fold increased risk (p=0.007) of MACE, this exceeding the risk of NYHA ≥II patients below this threshold. Separate NYHA I and ≥II sub-group multivariable models showed LV Mass z-score >3.9 to remain a strong and independent predictor of MACE, providing respective hazards of 2.6 and 4.7 (p=0.001 and <0.001). Conclusions NYHA-I patients with HCM experience significantly lower rates of MACE versus NYHA≥II. However, LV mass z-scoring permits powerful sub-stratification of these sub-groups, identifying NYHA-I patients with higher LV mass who will experience worse outcomes versus NYHA≥II patients with lower LV mass. This simple yet powerful stratification offers unique potential to identify asymptomatic patients with HCM who may benefit from targeted therapeutics to reduce future MACE.

Health Impact Assessment in Personalized Prevention: three applications on pharmacogenomic testing

Abstract Background In recent years, literature on the efficacy of genetic and genomic tests has expanded, yet the impact of personalized prevention programs that include such tests remains understudied. Our study aims to test the use of the Health Impact Assessment (HIA) framework, traditionally applied in non-health policy evaluation, to assess the impacts of personalized prevention policies. Methods We selected a policy mandating dihydropyrimidine dehydrogenase (DPYD) genetic test use before prescribing fluoropyrimidine for colorectal cancer (CRC) patients. Three HIAs are being conducted in Italy, Portugal, and Finland to assess the applicability across different contexts. The assessment involves literature review, comparative risk modeling, and stakeholder interviews. Three national steering committees, comprising representatives from key interest groups and experts, will evaluate evidence to gauge policy impacts and offer recommendations. Results The potential impacts identified include CRC patients’ outcomes, economic, organizational, patient experience, clinician workload, and equity. Preliminary results show a consensus in literature and national guidelines on the ability of the DPYD test to reduce adverse events and enhance the quality of life of CRC patients. However, the assessment reveals context-specific effects regarding equity, influenced by the prevalence of ethnic groups and variations in access to health services across regions. Differences also emerged on organizational aspects like the availability of resources such as laboratories and personnel. A lack of literature was observed on patient acceptability and clinician attitudes towards the test. Conclusions The use of HIA in personalized prevention might assist in identifying policy impacts often overlooked through the involvement of key interest groups in the assessment process. Such effort can further emphasize the significance of evaluating these programs while considering the geographical context. Key messages • Our study innovatively applies the Health Impact Assessment framework to evaluate personalized prevention programs in pharmacogenomic, addressing a critical research gap. • Preliminary findings underscore the importance of considering geographical factors in public health policy evaluation, highlighting nuanced context-specific effects.

Strategic Navigation of Firm-Specific Risks: A Comprehensive Analysis of Corporate Finance Challenges and Solutions

Abstract: This paper delves into the multifaceted nature of firm-specific risks (FSRs) within corporate finance, highlighting operational, financial, and strategic risks as pivotal challenges confronting companies across industries. Through detailed examinations of scenarios ranging from regulatory hurdles in the pharmaceutical industry to cyber threats in retail and currency fluctuations impacting multinational corporations, this study elucidates the intricate web of internal and external factors contributing to FSR. Furthermore, it explores the critical role of corporate governance, emphasizing board structure, transparency in financial reporting, and robust internal controls in effectively managing these risks. By integrating advanced data analytics and risk modeling with comprehensive risk management frameworks, the paper showcases strategies for identifying, assessing, and mitigating FSRs, thereby enhancing corporate resilience and strategic agility. This analysis not only contributes to the academic discourse on risk management but also offers practical insights for practitioners aiming to fortify their organizations against the unpredictability of the business landscape.

Experimental infection of ringtail possums (Pseudocheirus peregrinus) with Mycobacterium ulcerans, the agent of Buruli ulcer

Buruli ulcer (BU) is a necrotizing disease of skin and soft tissue caused by the bacterium Mycobacterium ulcerans (MU). In Australia, where the disease is emerging in new geographic areas and human case numbers are increasing, native possum species act as reservoir hosts. To better understand the life history of MU in one of its natural hosts, we conducted intra-dermal challenge of six wild caught, MU-naïve common ringtail possums (Pseudocheirus peregrinus). All six animals developed BU disease consistent with that observed in naturally infected ringtail possums. Time to ulceration varied between 49 and 77 days (mean = 61.8 days). Molecular evidence of systemic infection was detected in five animals and was supported by consistent histopathological findings in four animals. Pathological findings included random, multifocal, granulomatous hepatitis in four possums, one of which also had a mild, multifocal, interstitial granulomatous pneumonia. Acid-fast bacilli were only evident in inflammatory foci beyond the primary inoculation site in one possum. The ringtail possum model of MU infection is an important tool for the investigation of bacterial transmission dynamics, pathogenesis and immune response in a natural host. Data from this model may improve disease risk modelling and help identify intervention points to stop zoonotic transmission and disease spread.

Surrogate Modeling of Brain Trauma Risks Associated With Soccer Ball Headers

Abstract This study presents a coupled mathematical surrogate modeling and finite element analysis to quantify the risk of brain injury resulting from contact between a soccer ball and a player's head. Our approach implements a previously developed and validated regulation size finite element soccer ball and a 50th percentile adult male head model to inform the surrogate models regarding the positioning of the soccer ball's impact, inbound velocity, and the role of the activation of the neck musculature. Surrogate models are developed using injury metrics such as HIC15 and BrIC, which can be correlated to clinical assessments. The surrogate models predict occipital-temporal impacts as the highest risk of injury along a perpendicular axis and highlight the risk of increased velocity and passive neck musculature. The surrogate model predicts the worst-case scenarios to produce HIC15 and BrIC values of approximately 80 and 0.65, which correlate to 4.27% of a minor head injury (AIS1) and 71% of a moderate head injury (AIS2), respectively. These models are also used to develop a graphical user interface that may be used to supplement existing clinical knowledge of soccer headings to improve the head health of athletes participating in the game of soccer.

Airspace Constrained Free-Flight Analysis: Implications for Uncrewed Air Traffic Management

This paper provides a study of free-flight air traffic behaviour in increasingly constrained airspace environments. Traffic assumes three different free-flight operational constructs with airspace constraints considered as restricted (no-fly) regions. Simulations combine path planning and Monte Carlo techniques to qualitatively analyse emergent traffic behaviour and quantitatively assess spatial–temporal airspace conflict as the airspace constraints vary. Findings indicate that airspace constraints have a much stronger influence on aircraft behaviour than the free-flight operational construct, with any benefits of free flight rapidly diminishing as the airspace becomes more constrained. We conclude that structured traffic route (or network) designs and associated risk modelling approaches should be considered for safe and efficient traffic management of highly constrained and congested (or dense) airspace. This work therefore provides evidence to inform new airspace design and management initiatives, including low-altitude uncrewed traffic.

Predictive Modeling of Preeclampsia Risk Using Random Forest Algorithm within a Machine Learning Framework

Preeclampsia is a serious pregnancy complication characterized by high blood pressure, potentially leading to organ damage, making early risk prediction crucial to reducing maternal morbidity and mortality. This study aims to develop a preeclampsia risk prediction model using medical and clinical data from 80 patients at Rumah Bersalin Sadan. The data include demographic profiles, blood pressure, weight, maternal age, preeclampsia history, body mass index, number of previous pregnancies, as well as genetic and environmental factors. The dependent variable is the risk of preeclampsia, either as a binary outcome (yes/no) or as a continuous risk score. The predictive model was built using multivariate linear regression and the Random Forest algorithm. The results showed that the Random Forest model achieved an accuracy of 65.22%, with an F-statistic of 7.345 and a very small p-value (1.908e-06), indicating that the model effectively explains data variability. However, the low Kappa value suggests room for improvement through feature refinement, hyperparameter tuning, or exploring other algorithms. Although these findings suggest that Random Forest is a promising method, further evaluation and model optimization are needed to enhance predictive performance and determine whether this method is the most suitable for the dataset used.

Competing and Noncompeting Risk Models for Predicting Kidney Allograft Failure.

Prognostic models are becoming increasingly relevant in clinical trials as potential surrogate endpoints, and for patient management as clinical decision support tools. However, the impact of competing risks on model performance remains poorly investigated. We aimed to carefully assess the performance of competing risk and noncompeting risk models in the context of kidney transplantation, where allograft failure and death with a functioning graft are two competing outcomes. We included 11,046 kidney transplant recipients enrolled in 10 countries. We developed prediction models for long-term kidney graft failure prediction, without accounting (i.e., censoring) and accounting for the competing risk of death with a functioning graft, using Cox, Fine-Gray, and cause-specific Cox regression models. To this aim, we followed a detailed and transparent analytical framework for competing and noncompeting risk modelling, and carefully assessed the models' development, stability, discrimination, calibration, overall fit, clinical utility, and generalizability in external validation cohorts and subpopulations. More than 15 metrics were used to provide an exhaustive assessment of model performance. Among 11,046 recipients in the derivation and validation cohorts, 1,497 (14%) lost their graft and 1,003 (9%) died with a functioning graft after a median follow-up post-risk evaluation of 4.7 years (IQR 2.7-7.0). The cumulative incidence of graft loss was similarly estimated by Kaplan-Meier and Aalen-Johansen methods (17% versus 16% in the derivation cohort). Cox and competing risk models showed similar and stable risk estimates for predicting long-term graft failure (average mean absolute prediction error of 0.0140, 0.0138 and 0.0135 for Cox, Fine-Gray, and cause-specific Cox models, respectively). Discrimination and overall fit were comparable in the validation cohorts, with concordance index ranging from 0.76 to 0.87. Across various subpopulations and clinical scenarios, the models performed well and similarly, although in some high-risk groups (such as donors over 65 years old), the findings suggest a trend towards moderately improved calibration when using a competing risk approach. Competing and noncompeting risk models performed similarly in predicting long-term kidney graft failure.

The risk assessment of public-private partnership projects using artificial intelligence algorithms

Purpose: is to develop an innovative approach to risk management in public-private partnership (PPP) projects using advanced artificial intelligence technologies that allow creating the risk assessment model that takes into account non-linear relationships between various risk factors.Methods: in addition to traditional methods of scientific knowledge, interdisciplinary approaches of risk management and established practice of machine learning were used in the work. The methodological basis of the study was formed by works on the risk assessment and the application of AI algorithms in this area. The empirical basis of the study was the data of the official portal of ROSINFRA on public-private partnership projects.Results: the practice of applying AI algorithms to the task of assessing the risks of PPP-projects in Russia and abroad was studied. It is established that the most effective result is shown by the models based on Random-Forest-Classifier. However, the presented solutions do not take into account Russian economic realities. The authors have structured a database of implemented PPP-projects suitable for risk modeling. The model for assessing the risk of failure to achieve the objectives of Russian PPP-projects has been developed and its quality has been assessed. Recommendations on implementation of the model in the operational loop of PPP projects realization processes are offered.Conclusions and Relevance: the developed model allows, according to the general parameters of PPP project (region, authority, term of agreement, industry and scope of implementation, etc.) with the accuracy of 93% (according to the ROC\AUC metric), to assess the risk that the project will end incorrectly (due to a failed tender, refusal to launch, termination by court decision, cancellation/annulment of the tender). With the help of the model the executive authorities of the Russian Federation can build risk management for PPP projects management in the regions and thus contribute to their efficiency improvement. The article may also become useful for project management practitioners and appraisers.

FS900284 Survey of the Microbiological Contamination of Cull Ewes and Prime Lamb at Slaughter in England and Wales

The aim of this FSA-funded survey was to determine microbial contamination on sheep carcases in abattoirs in England and Wales as the last survey of healthy sheep at slaughter in England and Wales was undertaken 20 years ago. This project was run in conjunction with surveys funded by Defra and the PATH-SAFE research programme aimed at determining prevalence of caecal carriage of pathogens and levels of antimicrobial resistance in sheep at slaughter. Twelve out of 132 abattoirs in England and Wales were recruited, covering 28% of the annual national sheep slaughterhouse throughput. In total,1282 carcase swab samples were collected over a 12-month period from February 2023 to January 2024. All samples were tested for the isolation and quantification of Salmonella, 366 were tested for Campylobacter and 103 were tested for Enterobacteriaceae and E. coli. Salmonella was only detected in 7 (0.55%) of carcase swabs. Five isolates were confirmed as S. enterica subsp. diarizonae serotype 61:k:1:5, a sheep-adapted type with no confirmed association with human disease. The other two isolates were confirmed as S. Typhimurium. Campylobacter contamination was detected on 76 (20.8%) of carcases tested. Prevalence of E. coli (88.4%) and Enterobacteriaceae (75%) was as expected in freshly slaughtered meat, with 81.6% and 84.5% of these samples, respectively, presenting low levels of contamination The contamination detected on sheep carcases was significantly lower than that found in caecal samples examined in the Defra study suggesting that slaughtering processes are effective in reducing carcases being contaminated with pathogens. Correct cooking and hygienic handling and preparation of sheep meat will reduce/eliminate surface microbiological contamination further and reduce the risk to consumers. These results provide valuable baseline data that will help inform risk assessments and risk modelling and will inform future decisions by Government and the meat industry.

A conflict risk graph approach to modeling spatio-temporal dynamics of intersection safety

Intersections are among the most hazardous roadway spaces due to the complex and conflicting road users’ movements. Spatio-temporal modeling of conflict risks among road users can help to identify strategies to mitigate the exacerbation of safety risks and restore hazardous conditions to normal traffic situations. This paper proposes the 'Conflict Risk Graph' as a novel concept to infer real-time conflict risks at intersections at a fine-grained level by mapping conflict-prone locations to nodes within a network characterized by specific topological structures. A significant contribution of this work is the development of a Transformer-based Graph Convolutional Network (Trans-GCN), a model that synergistically combines the Transformer's proficiency in capturing global dependence with the GCN's ability to learn local correlations. The Trans-GCN adeptly models the complex evolution patterns of conflict risks at intersections. The evaluation in this paper against five common state-of-the-art deep learning approaches demonstrates the superior performance of the Trans-GCN in conflict risk inference and adaptability to node changes. Furthermore, extensive experiments with different node configurations reveal a correlation between node setup and model performance, showing that higher spatio-temporal resolution decreases inference accuracy. This insight informs the selection of an optimal node configuration that balances the detailed capture of spatio-temporal dynamics with modeling accuracy, enabling ideal conflict risk inferences at intersections. Ultimately, this work offers significant insights for the enhancement of proactive traffic safety management and the advancement of intelligent traffic systems.

Physical Vulnerability and Landslide Risk Assessment in Tegucigalpa City, Honduras

Quantitative disaster risk studies for slow-moving rotational and translational landslides in small regions (e.g., cities and watersheds) are very scarce. The limitations of risk modeling associated with these hazards include (i) the lack of data for physical modeling, (ii) methodological restrictions on estimating landslide intensity with statistical models and determining the temporal probability of landslides, and (iii) the absence of characterizations of the physical vulnerability of exposed assets. The present study combines and updates different methodologies to overcome these limitations for quantitative landslide disaster risk estimation, creating a novel methodological approach that was applied in a pilot study in Tegucigalpa city, Honduras. Tegucigalpa, the capital city of Honduras, has the highest number of recorded landslides in the country. In a previous study, landslides were found to be mainly concentrated in areas with colluvium and residual soils. As an input for the disaster risk assessment, this study generated landslide risk vulnerability functions based on empirical data. The application of the proposed methodology allowed us to estimate the average annual loss (AAL) caused by landslides in the study area—a key disaster risk metric that is lacking in other landslide disaster risk studies—enabling comparisons with disaster risk estimates associated with other hazards. In particular, the AAL value obtained for the study region was USD 7.26 million.

Projected risk and vulnerability to heat waves for Montreal, Quebec, using Gaussian processes

Urban areas face increasing climate risks and are at the forefront of adaptation challenges. Despite the growing number of cities that are developing adaptation plans, they often fail to implement, monitor, and evaluate them. This article addresses this issue by modelling a comprehensive risk assessment that includes vulnerability using Gaussian processes. Mortality during heat waves for the City of Montreal, Quebec, is used as a case study. The vulnerability model includes sensitivity components (age and socioeconomic variables) and an adaptive capacity component (a suitable level of vegetation to decrease the urban heat island effect). Various aging and climate scenarios (SSP1-2.6, SSP2-4.5 and SSP5-8.5) are used for projections up to year 2100. SHAP values are used to show features contributions to the model. As the climate warms, Montreal will face increasing summer mortality. The city should therefore increase its vegetation cover in vulnerable neighbourhoods. Despite inherent limitations to the complexity of risk modelling, this approach facilitates the implementation of adaptation solutions and their monitoring. Greater effort should be made in the future to improve comprehensive risk modelling and more research is required to validate which framework is best in closing the gap between science and political decisions.