Periodic Reporting for period 4 - EuroPOND (Data-driven models for Progression Of Neurological Disease)
Période du rapport: 2020-01-01 au 2020-12-31
EuroPOND developed uniquely powerful data-driven statistical-and-computational models of neurological disease progression. The models represent fundamentally new methods for understanding the complexity of the clinical presentation of neurological diseases, and the underlying biology. We demonstrated the models by generating new knowledge in a variety of neurological applications. Further, we explored avenues for patient benefit by using the models to underpin prototype support systems for clinical and drug-development applications. Specifically, the models and prototypes enable “Precision Medicine” by providing differential and personalised diagnosis, fine-grained staging, and personalised prognosis. These advances will contribute to positioning Europe as world leaders in one of the biggest challenges facing 21st century healthcare: management of neurological disease. Our specific objectives were: A) Develop, and implement as a prototype software tool, a new data-driven computational-and-statistical modelling framework for disease progression. B) Construct a quantitative validation framework using well-phenotyped and other complementary data sets from across the European arena. C) Construct models for a range of neurological conditions, specifically dementia, multiple sclerosis, prion diseases, and normal and abnormal neurodevelopment and ageing. D) Prototype a computer-assisted diagnosis and staging software system for dementia and demonstrate in relevant environments.
Key work during the first reporting period:
– Data collation/analysis plans (& best practice guidelines) in each neurological application
– Developed a suite of new state of the art data-driven disease progression modelling tools, including simulation framework for validation
– Prototyped a unified modelling framework
– Prototyped a software system for clinical use (dementia)
– Normal ageing models to inform disease models
– First models of disease progression in multiple sclerosis (MS), prion disease, and neurodevelopment
Key work during the second reporting period:
– Open-source software toolbox
– Enhancement of models: computational and technical
– Extensive evaluation of our model enhancements
– New models of MS progression
– Organised TADPOLE Challenge for forecasting Alzheimer’s: unique prospective design; £30K prize money; encouraged participation with live public webinars, open-source code, and an online community.
Key results during the third reporting period:
– Evaluated TADPOLE submissions (~90 worldwide). Allocated prizes. Marinescu, et al., (2020)
– Methodological updates extending capability of event-based models to handle high-dimensional imaging data (enabling automatic feature discovery)
– Characterised effects of fixed variables on Alzheimer’s progression and normal ageing
– Ageing-informed model for improved predictive inference in Alzheimer’s
– New model-based subtypes of MS, demonstrating utility for improving clinical trials: Eshaghi et al., Nature Communications (2021)
– New insight into prion diseases by revealing and characterising previously unknown subtype-specific differences in the spread of lesions
Key results during the fourth reporting period:
– Delivered an international workshop: CompAge2020 Computational approaches for ageing and age-related diseases. Content freely available at europond.github.io/compage2020
– New model-based subtypes of Alzheimer’s pathology: an international collaboration accepted in Vogel et al., Nature Medicine (2021). Potentially redefines the disease staging paradigm to in vivo (previously post-mortem) and to include subtyping. Provides mechanistic insight to inform drug development.
– Tested our clinical prototype with neurologists on real-world hospital data.
– Prototype software system for drug development in dementia, and for MS.
– Demonstrated translational utility of our models by training on research data and testing on hospital data (dementia).
– Delivered multiple data-driven models (and a web app) of normal ageing that characterise, for the first time, fixed-variable effects, model-based subtypes, and prediction of dementia cases in an asymptomatic population.
– Invented topological profiles of neurological disease, a characteristic combination of brain network descriptors that best describes pathology propagation. Provides new insight into biological mechanisms of Alzheimer’s, normal ageing, and MS.
– Delivered advanced models of prion disease progression that revealed important differences between traditional and model-based subtypes based on MRI, and generated mechanistic insight using spreading models.
– Delivered advanced models of neurodevelopment suggesting that: preterm birth causes a unique signature of both delayed and slowed brain growth; and that brain scans may not be a necessary component of patient management in preterm-born children and adolescents.
– Data collation/analysis plans (& best practice guidelines) in each neurological application
– Developed a suite of new state of the art data-driven disease progression modelling tools, including simulation framework for validation
– Prototyped a unified modelling framework
– Prototyped a software system for clinical use (dementia)
– Normal ageing models to inform disease models
– First models of disease progression in multiple sclerosis (MS), prion disease, and neurodevelopment
Key work during the second reporting period:
– Open-source software toolbox
– Enhancement of models: computational and technical
– Extensive evaluation of our model enhancements
– New models of MS progression
– Organised TADPOLE Challenge for forecasting Alzheimer’s: unique prospective design; £30K prize money; encouraged participation with live public webinars, open-source code, and an online community.
Key results during the third reporting period:
– Evaluated TADPOLE submissions (~90 worldwide). Allocated prizes. Marinescu, et al., (2020)
– Methodological updates extending capability of event-based models to handle high-dimensional imaging data (enabling automatic feature discovery)
– Characterised effects of fixed variables on Alzheimer’s progression and normal ageing
– Ageing-informed model for improved predictive inference in Alzheimer’s
– New model-based subtypes of MS, demonstrating utility for improving clinical trials: Eshaghi et al., Nature Communications (2021)
– New insight into prion diseases by revealing and characterising previously unknown subtype-specific differences in the spread of lesions
Key results during the fourth reporting period:
– Delivered an international workshop: CompAge2020 Computational approaches for ageing and age-related diseases. Content freely available at europond.github.io/compage2020
– New model-based subtypes of Alzheimer’s pathology: an international collaboration accepted in Vogel et al., Nature Medicine (2021). Potentially redefines the disease staging paradigm to in vivo (previously post-mortem) and to include subtyping. Provides mechanistic insight to inform drug development.
– Tested our clinical prototype with neurologists on real-world hospital data.
– Prototype software system for drug development in dementia, and for MS.
– Demonstrated translational utility of our models by training on research data and testing on hospital data (dementia).
– Delivered multiple data-driven models (and a web app) of normal ageing that characterise, for the first time, fixed-variable effects, model-based subtypes, and prediction of dementia cases in an asymptomatic population.
– Invented topological profiles of neurological disease, a characteristic combination of brain network descriptors that best describes pathology propagation. Provides new insight into biological mechanisms of Alzheimer’s, normal ageing, and MS.
– Delivered advanced models of prion disease progression that revealed important differences between traditional and model-based subtypes based on MRI, and generated mechanistic insight using spreading models.
– Delivered advanced models of neurodevelopment suggesting that: preterm birth causes a unique signature of both delayed and slowed brain growth; and that brain scans may not be a necessary component of patient management in preterm-born children and adolescents.
Management and treatment of neurological diseases is one of our biggest medical challenges, with increasingly devastating socioeconomic outcomes as the global population ages. The current paradigm is knowledge-driven clinical decisions, but the snowstorm of data available defies qualitative clinical evaluation; the heterogeneity of data types complicates use of traditional statistical methods; and the large datasets available remain far from the big-data sizes necessary for fully data-driven machine-learning.
EuroPOND aimed to replace the current paradigm with data-driven disease progression modelling, which strikes a balance between knowledge-driven and data-driven approaches. We advanced the state of the art by developing new models predictive of disease progression in individual patients and informative of underlying biological patterns. We demonstrated utility by prototyping clinical and drug-development tools.
We contributed to success of future actions: 1) producing best practice guidelines for performing and validating such collaborative medical research at scale; 2) releasing our advanced models as an open-source resource; and 3) initiating multiple follow-on funding successes, e.g. E-DADS, I-AIM, TADPOLE-SHARE.
Key impactful discoveries include paradigm-shifting subtyping results in dementia and MS: our data-driven subgroups of dementia are used by researchers to frame their thinking on how to setup new clinical studies; our data-driven subtyping in MS has redefined disease understanding, with huge ramifications for patient management and treatment assignment in clinical trials.
We continue to run TADPOLE Challenge to forecast Alzheimer’s progression. The TADPOLE data and evaluation metrics have become a standard way to evaluate new algorithms and our challenge winners provided a new, previously unavailable, benchmark and standard for the field (media: ALZFORUM 1, 2).
Our prototype clinical tool was tested extensively by multiple clinical user groups and is deployed as part of icobrain dm, which includes CE/FDA-approved MRI analysis and state of the art data protection. This is expected to impact hospitals and clinical research organisations across Europe.
Conclusion: In combination, we expect broad impact of EuroPOND's work in facilitating therapy development and deployment through earlier and discriminative diagnosis, as well as clear and consistent staging and stratification. This will ultimately benefit society by improving quality of life and reducing loss of productivity for individuals, families, healthcare workers, and carers.
EuroPOND aimed to replace the current paradigm with data-driven disease progression modelling, which strikes a balance between knowledge-driven and data-driven approaches. We advanced the state of the art by developing new models predictive of disease progression in individual patients and informative of underlying biological patterns. We demonstrated utility by prototyping clinical and drug-development tools.
We contributed to success of future actions: 1) producing best practice guidelines for performing and validating such collaborative medical research at scale; 2) releasing our advanced models as an open-source resource; and 3) initiating multiple follow-on funding successes, e.g. E-DADS, I-AIM, TADPOLE-SHARE.
Key impactful discoveries include paradigm-shifting subtyping results in dementia and MS: our data-driven subgroups of dementia are used by researchers to frame their thinking on how to setup new clinical studies; our data-driven subtyping in MS has redefined disease understanding, with huge ramifications for patient management and treatment assignment in clinical trials.
We continue to run TADPOLE Challenge to forecast Alzheimer’s progression. The TADPOLE data and evaluation metrics have become a standard way to evaluate new algorithms and our challenge winners provided a new, previously unavailable, benchmark and standard for the field (media: ALZFORUM 1, 2).
Our prototype clinical tool was tested extensively by multiple clinical user groups and is deployed as part of icobrain dm, which includes CE/FDA-approved MRI analysis and state of the art data protection. This is expected to impact hospitals and clinical research organisations across Europe.
Conclusion: In combination, we expect broad impact of EuroPOND's work in facilitating therapy development and deployment through earlier and discriminative diagnosis, as well as clear and consistent staging and stratification. This will ultimately benefit society by improving quality of life and reducing loss of productivity for individuals, families, healthcare workers, and carers.