Hamish Patten
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Research Subjects
Hamish Patten has a total of 110 co-authors. Here are the top co-authors by number of shared publications:
- Jonathan P. Graves
École Polytechnique Fédérale de Lausanne
4 shared publications - David Pfefferlé
The University of Western Australia, 35 Stirling Highway, Crawley, WA 6009, Australia
3 shared publications David Steinsaltz
Department of Statistics, University of Oxford, Oxford, United Kingdom
3 shared publicationsJet Contributors
3 shared publications- Andreas Kleiner
Princeton Plasma Physics Laboratory, Princeton, NJ, United States of America
2 shared publications - David Steinsaltz
Department of Statistics, University of Oxford, Oxford OX1 4BH, United Kingdom
2 shared publications
Latest Publications
- Open Access
- 21 August 2026
- Published by Oxford University Press (OUP) in Journal of the Royal Statistical Society Series C: Applied Statistics
A Bayesian approach for earthquake impact modelling
Immediately following a disaster event, such as an earthquake, estimates of the damage extent play a key role in informing the coordination of response and recovery efforts. We develop a novel impact estimation tool that leverages a generalized Bayesian approach to generate earthquake impact estimates across three impact types: mortality, population displacement, and building damage. Inference is performed within a likelihood-free framework, and a scoring-rule-based posterior avoids information loss from non-sufficient summary statistics. We propose an adaptation of existing scoring-rule-based loss functions that accommodates the use of an approximate Bayesian computation sequential Monte Carlo framework. The model achieves improved results compared to two leading impact estimation tools in the prediction of total mortality when tested on a set of held-out past events. The proposed method provides four further advantages over existing empirical approaches: modelling produces a gridded spatial map of the estimated impact, predictions benefit from the Bayesian quantification and interpretation of uncertainty, there is direct handling of multi-shock earthquake events, and the use of a joint model between impact types allows predictions to be updated as impact observations become available.
- 21 August 2026
- Published by Oxford University Press (OUP) in Journal of the Royal Statistical Society Series C: Applied Statistics
Open AccessA Bayesian approach for earthquake impact modelling
Immediately following a disaster event, such as an earthquake, estimates of the damage extent play a key role in informing the coordination of response and recovery efforts. We develop a novel impact estimation tool that leverages a generalized Bayesian approach to generate earthquake impact estimates across three impact types: mortality, population displacement, and building damage. Inference is performed within a likelihood-free framework, and a scoring-rule-based posterior avoids information loss from non-sufficient summary statistics. We propose an adaptation of existing scoring-rule-based loss functions that accommodates the use of an approximate Bayesian computation sequential Monte Carlo framework. The model achieves improved results compared to two leading impact estimation tools in the prediction of total mortality when tested on a set of held-out past events. The proposed method provides four further advantages over existing empirical approaches: modelling produces a gridded spatial map of the estimated impact, predictions benefit from the Bayesian quantification and interpretation of uncertainty, there is direct handling of multi-shock earthquake events, and the use of a joint model between impact types allows predictions to be updated as impact observations become available.
- Top Cited PapersOpen Access
- 24 March 2026
- Published by Oxford University Press (OUP) in Journal of the Royal Statistical Society Series A: Statistics in Society
Bayesian spatiotemporal modelling of political violence and conflict events using discrete-time Hawkes processes
The monitoring of conflict risk in the humanitarian sector is largely based on simple historic averages. The overarching goal of this work is to assess the potential for using a more statistically rigorous approach to monitor the risk of political violence and conflict events in practice, and thereby improve our understanding of their temporal and spatial patterns, to inform preventative measures. In particular, a Bayesian, spatiotemporal variant of the Hawkes process is fitted to data gathered by the Armed Conflict Location and Event Data (ACLED) project to obtain sub-national estimates of conflict risk in South Asia over time and space. Our model can effectively estimate the risk level of these events within a statistically sound framework, with a more precise understanding of uncertainty than was previously possible. The model also provides insights into differences in behaviours between countries and conflict types. We also show how our model can be used to monitor short and long term trends, and that it is more stable and robust to outliers compared to current practices that rely on historical averages.
- 24 March 2026
- Published by Oxford University Press (OUP) in Journal of the Royal Statistical Society Series A: Statistics in Society
Top Cited PapersOpen AccessBayesian spatiotemporal modelling of political violence and conflict events using discrete-time Hawkes processes
The monitoring of conflict risk in the humanitarian sector is largely based on simple historic averages. The overarching goal of this work is to assess the potential for using a more statistically rigorous approach to monitor the risk of political violence and conflict events in practice, and thereby improve our understanding of their temporal and spatial patterns, to inform preventative measures. In particular, a Bayesian, spatiotemporal variant of the Hawkes process is fitted to data gathered by the Armed Conflict Location and Event Data (ACLED) project to obtain sub-national estimates of conflict risk in South Asia over time and space. Our model can effectively estimate the risk level of these events within a statistically sound framework, with a more precise understanding of uncertainty than was previously possible. The model also provides insights into differences in behaviours between countries and conflict types. We also show how our model can be used to monitor short and long term trends, and that it is more stable and robust to outliers compared to current practices that rely on historical averages.
- Open Access
- 15 March 2025
- Published by Copernicus GmbH
Applying AI to Classify Hazards from Text
Classifying which hazards are present when recording disaster hazard and impact data is crucial in order to maximise the utility of the data in the future. For example, it is estimated that seven out of eight deaths in the US from tropical cyclones were not caused by excess wind speeds but due to hazards such as coastal and pluvial flooding. For actors such as National Disaster Management Offices (NDMOs) and Non-Governmental Organisations (NGOs), improving disaster response, preparedness and mitigation relies on understanding historical disaster hazards and their impacts. Contingency plans, for example, between riverine and pluvial flooding should differ, as the former can take days to build up whereas the latter can impact a local environment within an hour. However, with many databases simply labelling the disaster as a 'flood', understanding whether the risk in an area is predominantly one hazard more than another makes decision making and disaster risk analysis difficult.This research uses text descriptions of tens of thousands of disaster reports from sources such as IFRC Field Reports, ReliefWeb, EM-DAT and Twitter. By applying Large Language Models (LLMs) and Natural Language Processing (NLP) models, multi-hazard classification is made. The LLM, ALBERT, outperforms the other models, with the advantage of LLMs over NLPs that no pre-training of the model is required (zero-shot). We present and overview of these results and discuss how they may be useful to the disaster risk community.
- 15 March 2025
- Published by Copernicus GmbH
Open AccessApplying AI to Classify Hazards from Text
Classifying which hazards are present when recording disaster hazard and impact data is crucial in order to maximise the utility of the data in the future. For example, it is estimated that seven out of eight deaths in the US from tropical cyclones were not caused by excess wind speeds but due to hazards such as coastal and pluvial flooding. For actors such as National Disaster Management Offices (NDMOs) and Non-Governmental Organisations (NGOs), improving disaster response, preparedness and mitigation relies on understanding historical disaster hazards and their impacts. Contingency plans, for example, between riverine and pluvial flooding should differ, as the former can take days to build up whereas the latter can impact a local environment within an hour. However, with many databases simply labelling the disaster as a 'flood', understanding whether the risk in an area is predominantly one hazard more than another makes decision making and disaster risk analysis difficult.This research uses text descriptions of tens of thousands of disaster reports from sources such as IFRC Field Reports, ReliefWeb, EM-DAT and Twitter. By applying Large Language Models (LLMs) and Natural Language Processing (NLP) models, multi-hazard classification is made. The LLM, ALBERT, outperforms the other models, with the advantage of LLMs over NLPs that no pre-training of the model is required (zero-shot). We present and overview of these results and discuss how they may be useful to the disaster risk community.
- Open Access
- 22 October 2024
- Published by Elsevier in The American Journal of Cardiology
- Vol. 234, 71-78
Short-Term and Mid-Term Blood Pressure Variability and Long-Term Mortality
- 22 October 2024
- Published by Elsevier in The American Journal of Cardiology
- Vol. 234, 71-78
Open AccessShort-Term and Mid-Term Blood Pressure Variability and Long-Term Mortality
- Open Access
- 14 June 2024
- Published by Springer Nature in International Journal of Disaster Risk Science
- Vol. 15 (3) , 421-433
Data-Driven Earthquake Multi-impact Modeling: A Comparison of Models
In this study, a broad range of supervised machine learning and parametric statistical, geospatial, and non-geospatial models were applied to model both aggregated observed impact estimate data and satellite image-derived geolocated building damage data for earthquakes, via regression- and classification-based models, respectively. For the aggregated observational data, models were ranked via predictive performance of mortality, population displacement, building damage, and building destruction for 375 observations across 161 earthquakes in 61 countries. For the satellite image-derived data, models were ranked via classification performance (damaged/unaffected) of 369,813 geolocated buildings for 26 earthquakes in 15 countries. Grouped k-fold, 3-repeat cross validation was used to ensure out-of-sample predictive performance. Feature importance of several variables used as proxies for vulnerability to disasters indicates covariate utility. The 2023 Türkiye–Syria earthquake event was used to explore model limitations for extreme events. However, applying the AdaBoost model on the 27,032 held-out buildings of the 2023 Türkiye–Syria earthquake event, predictions had an AUC of 0.93. Therefore, without any geospatial, building-specific, or direct satellite image information, this model accurately classified building damage, with significantly improved performance over satellite image trained models found in the literature.
- 14 June 2024
- Published by Springer Nature in International Journal of Disaster Risk Science
- Vol. 15 (3) , 421-433
Open AccessData-Driven Earthquake Multi-impact Modeling: A Comparison of Models
In this study, a broad range of supervised machine learning and parametric statistical, geospatial, and non-geospatial models were applied to model both aggregated observed impact estimate data and satellite image-derived geolocated building damage data for earthquakes, via regression- and classification-based models, respectively. For the aggregated observational data, models were ranked via predictive performance of mortality, population displacement, building damage, and building destruction for 375 observations across 161 earthquakes in 61 countries. For the satellite image-derived data, models were ranked via classification performance (damaged/unaffected) of 369,813 geolocated buildings for 26 earthquakes in 15 countries. Grouped k-fold, 3-repeat cross validation was used to ensure out-of-sample predictive performance. Feature importance of several variables used as proxies for vulnerability to disasters indicates covariate utility. The 2023 Türkiye–Syria earthquake event was used to explore model limitations for extreme events. However, applying the AdaBoost model on the 27,032 held-out buildings of the 2023 Türkiye–Syria earthquake event, predictions had an AUC of 0.93. Therefore, without any geospatial, building-specific, or direct satellite image information, this model accurately classified building damage, with significantly improved performance over satellite image trained models found in the literature.
- Open Access
Short-term and Mid-term Blood Pressure Variability and long-term Mortality: evidence from the Third National Health and Nutrition Examination Study
In considering the impact of blood-pressure variability on health outcomes, two methodological challenges arise: The presence of multiple timescales of variability that may act independently and interactively, and the fairly large stochastic uncertainty that is inevitable in estimates of individual variability based on modest numbers of observations. Here we present an application of Bayesian hierarchical modeling to the problem of estimating the effect of blood pressure (BP) variability on all-cause and cardiovascular mortality with two timescales – short-term variation among multiple measures at one visit, and medium-term variation between the measures at two visits several months apart. We use data from the Third National Health and Nutrition Examination Survey linked with up to 27 years of mortality follow-up. We find that medium-term systolic BP variability had a very significant predictive value for CV and all-cause mortality, around one-third as large as the well-established impact of mean systolic BP. Medium-term diastolic variability had an additional, though smaller, predictive effect. Short-term variability, on the other hand, had little or no measurable predictive value. The medium-term variability effect persisted when controlling for Framingham risk score.
Open AccessShort-term and Mid-term Blood Pressure Variability and long-term Mortality: evidence from the Third National Health and Nutrition Examination Study
In considering the impact of blood-pressure variability on health outcomes, two methodological challenges arise: The presence of multiple timescales of variability that may act independently and interactively, and the fairly large stochastic uncertainty that is inevitable in estimates of individual variability based on modest numbers of observations. Here we present an application of Bayesian hierarchical modeling to the problem of estimating the effect of blood pressure (BP) variability on all-cause and cardiovascular mortality with two timescales – short-term variation among multiple measures at one visit, and medium-term variation between the measures at two visits several months apart. We use data from the Third National Health and Nutrition Examination Survey linked with up to 27 years of mortality follow-up. We find that medium-term systolic BP variability had a very significant predictive value for CV and all-cause mortality, around one-third as large as the well-established impact of mean systolic BP. Medium-term diastolic variability had an additional, though smaller, predictive effect. Short-term variability, on the other hand, had little or no measurable predictive value. The medium-term variability effect persisted when controlling for Framingham risk score.
- Open Access
- 19 July 2023
- Published by Springer Nature in Research Square
Data-Driven Earthquake Multi-Impact Modelling: A Comparison of Models
A broad range of supervised Machine Learning and parametric statistical, geospatial and non-geospatial models are applied to model both aggregated observed impact estimate count data and satellite image-derived geo-located building damage labelled data, via regression- and classification-based models, respectively. For the former, models are ranked via predictive performance of mortality, population displacement, building damage and building destruction for 161 earthquakes in 61 countries. For the latter, models were ranked via classification performance (damaged/unaffected) of 369,813 geo-located buildings for 26 earthquakes in 15 countries. k-fold, 3-repeat cross validation is utilised to ensure out-of-sample predictive performance. Feature importance on several variables used as proxies for disaster-risk vulnerability indicates covariate utility. The 2023 Türkiye-Syria earthquakes event is shown to be unlike any of the other 161 earthquakes, resulting in high prediction error for the aggregated impact models compared to all other events, reflecting model limitations on extreme events. However, training the building damage classifier AdaBoost model on all except the 2023 Türkiye-Syria earthquakes, predictions on the 27,032 held-out buildings have an AUC of 0.93. Therefore, without any geospatial, building-specific or direct satellite image information, this model accurately classifies building damage, with significantly improved performance over satellite-image trained models found in the literature.
- 19 July 2023
- Published by Springer Nature in Research Square
Open AccessData-Driven Earthquake Multi-Impact Modelling: A Comparison of Models
A broad range of supervised Machine Learning and parametric statistical, geospatial and non-geospatial models are applied to model both aggregated observed impact estimate count data and satellite image-derived geo-located building damage labelled data, via regression- and classification-based models, respectively. For the former, models are ranked via predictive performance of mortality, population displacement, building damage and building destruction for 161 earthquakes in 61 countries. For the latter, models were ranked via classification performance (damaged/unaffected) of 369,813 geo-located buildings for 26 earthquakes in 15 countries. k-fold, 3-repeat cross validation is utilised to ensure out-of-sample predictive performance. Feature importance on several variables used as proxies for disaster-risk vulnerability indicates covariate utility. The 2023 Türkiye-Syria earthquakes event is shown to be unlike any of the other 161 earthquakes, resulting in high prediction error for the aggregated impact models compared to all other events, reflecting model limitations on extreme events. However, training the building damage classifier AdaBoost model on all except the 2023 Türkiye-Syria earthquakes, predictions on the 27,032 held-out buildings have an AUC of 0.93. Therefore, without any geospatial, building-specific or direct satellite image information, this model accurately classifies building damage, with significantly improved performance over satellite-image trained models found in the literature.
- Open Access
- 27 April 2023
- Published by Cambridge University Press (CUP) in Journal of Plasma Physics
- Vol. 89 (2) , 955890202
Fast ion generation by combined RF-NBI heating in W7-X
Good fast ion confinement in quasi-isodynamic stellarators still has to be demonstrated experimentally. In the absence of fusion alphas, auxiliary heating systems can be used to generate fast ions. For this purpose, W7-X has neutral beam injection and its newly installed ion cyclotron heating system. In practice, both systems can run simultaneously, the synergetic running of both is the focus of the current modelling development. With the upgrades to the SCENIC code described here, we find that, compared with pure neutral beam injection (NBI), the fast ion tail is significantly enhanced when radio frequency (RF) heating is enabled. But the effectiveness of heating and fast ion tail generation strongly depends on the plasma parameters and antenna frequency, which calls for further study, both numerical and experimental.
- 27 April 2023
- Published by Cambridge University Press (CUP) in Journal of Plasma Physics
- Vol. 89 (2) , 955890202
Open AccessFast ion generation by combined RF-NBI heating in W7-X
Good fast ion confinement in quasi-isodynamic stellarators still has to be demonstrated experimentally. In the absence of fusion alphas, auxiliary heating systems can be used to generate fast ions. For this purpose, W7-X has neutral beam injection and its newly installed ion cyclotron heating system. In practice, both systems can run simultaneously, the synergetic running of both is the focus of the current modelling development. With the upgrades to the SCENIC code described here, we find that, compared with pure neutral beam injection (NBI), the fast ion tail is significantly enhanced when radio frequency (RF) heating is enabled. But the effectiveness of heating and fast ion tail generation strongly depends on the plasma parameters and antenna frequency, which calls for further study, both numerical and experimental.
- 26 January 2021
- Published by IOP Publishing in Nuclear Fusion
- Vol. 61 (4) , 046014
Ion cyclotron resonance heating with shifted separatrix
Ion cyclotron resonance heating, which we refer to as ICRH-SS, with the quasilinear (QL) separatrix (i.e., the separatrix in the space of QL routes of ion acceleration) located in the region of passing particles is studied. The aim of ICRH-SS is to minimize the fraction of trapped particles (particles with small longitudinal velocities) in the population of fast ions. The basic idea of ICRH-SS—shifting the separatrix to the region of passing particles—was advanced in the paper (Kolesnichenko et al 2017 Nuclear Fusion 57 066004). In this work, new features of ICRH-SS are revealed. The 3D QL routes of the particle acceleration and effects of Coulomb collisions are studied. A QL equation for distribution function of neutral beam injection (NBI) ions, which is convenient for analysis, is derived. Conditions for QL flux prevailing over collisional flux caused by pitch scattering are obtained. Numerical simulations using the SCENIC package are carried out for a JET plasma with NBI ions that are accelerated by RF field in the core region. A JET pulse designed as a demonstration of the so called ‘three ions’ scheme, which also complies with the criteria of ICRH-SS scheme, was selected. The numerical results show that in the considered example most accelerated ions have larger longitudinal velocities and fast particle orbits are passing during ICRH-SS, whereas, ‘conventional’ ICRH (defined as ICRH with the separatrix in the region of trapped particles) produces accelerated ions with banana orbits. Numerical results also show an increase in fast ion generation and core plasma heating performance for ICRH-SS as compared to the conventional ICRH.
- 26 January 2021
- Published by IOP Publishing in Nuclear Fusion
- Vol. 61 (4) , 046014
Ion cyclotron resonance heating with shifted separatrix
Ion cyclotron resonance heating, which we refer to as ICRH-SS, with the quasilinear (QL) separatrix (i.e., the separatrix in the space of QL routes of ion acceleration) located in the region of passing particles is studied. The aim of ICRH-SS is to minimize the fraction of trapped particles (particles with small longitudinal velocities) in the population of fast ions. The basic idea of ICRH-SS—shifting the separatrix to the region of passing particles—was advanced in the paper (Kolesnichenko et al 2017 Nuclear Fusion 57 066004). In this work, new features of ICRH-SS are revealed. The 3D QL routes of the particle acceleration and effects of Coulomb collisions are studied. A QL equation for distribution function of neutral beam injection (NBI) ions, which is convenient for analysis, is derived. Conditions for QL flux prevailing over collisional flux caused by pitch scattering are obtained. Numerical simulations using the SCENIC package are carried out for a JET plasma with NBI ions that are accelerated by RF field in the core region. A JET pulse designed as a demonstration of the so called ‘three ions’ scheme, which also complies with the criteria of ICRH-SS scheme, was selected. The numerical results show that in the considered example most accelerated ions have larger longitudinal velocities and fast particle orbits are passing during ICRH-SS, whereas, ‘conventional’ ICRH (defined as ICRH with the separatrix in the region of trapped particles) produces accelerated ions with banana orbits. Numerical results also show an increase in fast ion generation and core plasma heating performance for ICRH-SS as compared to the conventional ICRH.
- 22 May 2020
- Published by IOP Publishing in Nuclear Fusion
- Vol. 60 (6) , 066029
Predictive multi-channel flux-driven modelling to optimise ICRH tungsten control and fusion performance in JET
- 22 May 2020
- Published by IOP Publishing in Nuclear Fusion
- Vol. 60 (6) , 066029
Predictive multi-channel flux-driven modelling to optimise ICRH tungsten control and fusion performance in JET