×
  • Select the area you would like to search.
  • ACTIVE INVESTIGATIONS Search for current projects using the investigator's name, institution, or keywords.
  • EXPERTS KNOWLEDGE BASE Enter keywords to search a list of questions and answers received and processed by the ADNI team.
  • ADNI PDFS Search any ADNI publication pdf by author, keyword, or PMID. Use an asterisk only to view all pdfs.
Principal Investigator  
Principal Investigator's Name: Virginie Buggia-Prevot
Institution: Valo Health
Department: Neurology Discovery
Country:
Proposed Analysis: There is a significant need to develop disease modifying therapies (DMTs) for Alzheimer’s disease (AD), given that current therapeutic options provide minimal to modest symptomatic relief for patients. Historically, our ability to identify novel drug targets has been limited due challenges stratifying patient populations and the delay between diagnosis and intervention. We now have an opportunity to take advantage of rich data sources, like those provided through ADNI, to overcome these limitations and propose to do so by evaluating these data using machine learning. To meet the challenges associated with developing new DMTs for AD, we will build machine learning (ML) models which aim to accurately predict risk, timing of onset, and the course and kinetics of progression of disease mechanisms, neurodegeneration, and clinical symptoms across data-driven patient subpopulations based on clinical, laboratory, omic, and genetic data. Based on this understanding, we propose to further identify the mechanisms underlying patient worsening to the advanced stage of AD, which contribute to the need for hospice care and to premature death, to offer therapeutic options for the current highest unmet need. Several high-dimensional machine learning techniques that have shown initial promise in analyzing patient data will be applied to identify patient subtypes, such as denoising autoencoders and hierarchical latent bayesian methods to learn patient representations derived directly from ADNI patient data, paired with other large datasets such as AIBL patient data. These representations will be used to predict conversion to AD symptom and disease progression and to perform unsupervised clustering of patients, revealing patient subtypes reflective of distinct disease trajectories and etiologies. In order to better match patients to drugs and future drug targets, we will leverage recent advances in high-dimensional causal inference and counterfactual prediction. In particular, we will use representation learning approaches to estimate individual treatment effects and counterfactual Gaussian process approaches to simulate alternative drug target response trajectories for individual patients.
Additional Investigators  
Investigator's Name: Brant Peterson
Proposed Analysis: There is a significant need to develop disease modifying therapies (DMTs) for Alzheimer’s disease (AD), given that current therapeutic options provide minimal to modest symptomatic relief for patients. Historically, our ability to identify novel drug targets has been limited due challenges stratifying patient populations and the delay between diagnosis and intervention. We now have an opportunity to take advantage of rich data sources, like those provided through ADNI, to overcome these limitations and propose to do so by evaluating these data using machine learning. To meet the challenges associated with developing new DMTs for AD, we will build machine learning (ML) models which aim to accurately predict risk, timing of onset, and the course and kinetics of progression of disease mechanisms, neurodegeneration, and clinical symptoms across data-driven patient subpopulations based on clinical, laboratory, omic, and genetic data. Based on this understanding, we propose to further identify the mechanisms underlying patient worsening to the advanced stage of AD, which contribute to the need for hospice care and to premature death, to offer therapeutic options for the current highest unmet need. Several high-dimensional machine learning techniques that have shown initial promise in analyzing patient data will be applied to identify patient subtypes, such as denoising autoencoders and hierarchical latent bayesian methods to learn patient representations derived directly from ADNI patient data, paired with other large datasets such as AIBL patient data. These representations will be used to predict conversion to AD symptom and disease progression and to perform unsupervised clustering of patients, revealing patient subtypes reflective of distinct disease trajectories and etiologies. In order to better match patients to drugs and future drug targets, we will leverage recent advances in high-dimensional causal inference and counterfactual prediction. In particular, we will use representation learning approaches to estimate individual treatment effects and counterfactual Gaussian process approaches to simulate alternative drug target response trajectories for individual patients.
Investigator's Name: Derek Drake
Proposed Analysis: There is a significant need to develop disease modifying therapies (DMTs) for Alzheimer’s disease (AD), given that current therapeutic options provide minimal to modest symptomatic relief for patients. Historically, our ability to identify novel drug targets has been limited due challenges stratifying patient populations and the delay between diagnosis and intervention. We now have an opportunity to take advantage of rich data sources, like those provided through ADNI, to overcome these limitations and propose to do so by evaluating these data using machine learning. To meet the challenges associated with developing new DMTs for AD, we will build machine learning (ML) models which aim to accurately predict risk, timing of onset, and the course and kinetics of progression of disease mechanisms, neurodegeneration, and clinical symptoms across data-driven patient subpopulations based on clinical, laboratory, omic, and genetic data. Based on this understanding, we propose to further identify the mechanisms underlying patient worsening to the advanced stage of AD, which contribute to the need for hospice care and to premature death, to offer therapeutic options for the current highest unmet need. Several high-dimensional machine learning techniques that have shown initial promise in analyzing patient data will be applied to identify patient subtypes, such as denoising autoencoders and hierarchical latent bayesian methods to learn patient representations derived directly from ADNI patient data, paired with other large datasets such as AIBL patient data. These representations will be used to predict conversion to AD symptom and disease progression and to perform unsupervised clustering of patients, revealing patient subtypes reflective of distinct disease trajectories and etiologies. In order to better match patients to drugs and future drug targets, we will leverage recent advances in high-dimensional causal inference and counterfactual prediction. In particular, we will use representation learning approaches to estimate individual treatment effects and counterfactual Gaussian process approaches to simulate alternative drug target response trajectories for individual patients.
Investigator's Name: Jessica Sadick
Proposed Analysis: There is a significant need to develop disease modifying therapies (DMTs) for Alzheimer’s disease (AD), given that current therapeutic options provide minimal to modest symptomatic relief for patients. Historically, our ability to identify novel drug targets has been limited due challenges stratifying patient populations and the delay between diagnosis and intervention. We now have an opportunity to take advantage of rich data sources, like those provided through ADNI, to overcome these limitations and propose to do so by evaluating these data using machine learning. To meet the challenges associated with developing new DMTs for AD, we will build machine learning (ML) models which aim to accurately predict risk, timing of onset, and the course and kinetics of progression of disease mechanisms, neurodegeneration, and clinical symptoms across data-driven patient subpopulations based on clinical, laboratory, omic, and genetic data. Based on this understanding, we propose to further identify the mechanisms underlying patient worsening to the advanced stage of AD, which contribute to the need for hospice care and to premature death, to offer therapeutic options for the current highest unmet need. Several high-dimensional machine learning techniques that have shown initial promise in analyzing patient data will be applied to identify patient subtypes, such as denoising autoencoders and hierarchical latent bayesian methods to learn patient representations derived directly from ADNI patient data, paired with other large datasets such as AIBL patient data. These representations will be used to predict conversion to AD symptom and disease progression and to perform unsupervised clustering of patients, revealing patient subtypes reflective of distinct disease trajectories and etiologies. In order to better match patients to drugs and future drug targets, we will leverage recent advances in high-dimensional causal inference and counterfactual prediction. In particular, we will use representation learning approaches to estimate individual treatment effects and counterfactual Gaussian process approaches to simulate alternative drug target response trajectories for individual patients.
Investigator's Name: Alexander Rajan
Proposed Analysis: There is a significant need to develop disease modifying therapies (DMTs) for Alzheimer’s disease (AD), given that current therapeutic options provide minimal to modest symptomatic relief for patients. Historically, our ability to identify novel drug targets has been limited due challenges stratifying patient populations and the delay between diagnosis and intervention. We now have an opportunity to take advantage of rich data sources, like those provided through ADNI, to overcome these limitations and propose to do so by evaluating these data using machine learning. To meet the challenges associated with developing new DMTs for AD, we will build machine learning (ML) models which aim to accurately predict risk, timing of onset, and the course and kinetics of progression of disease mechanisms, neurodegeneration, and clinical symptoms across data-driven patient subpopulations based on clinical, laboratory, omic, and genetic data. Based on this understanding, we propose to further identify the mechanisms underlying patient worsening to the advanced stage of AD, which contribute to the need for hospice care and to premature death, to offer therapeutic options for the current highest unmet need. Several high-dimensional machine learning techniques that have shown initial promise in analyzing patient data will be applied to identify patient subtypes, such as denoising autoencoders and hierarchical latent bayesian methods to learn patient representations derived directly from ADNI patient data, paired with other large datasets such as AIBL patient data. These representations will be used to predict conversion to AD symptom and disease progression and to perform unsupervised clustering of patients, revealing patient subtypes reflective of distinct disease trajectories and etiologies. In order to better match patients to drugs and future drug targets, we will leverage recent advances in high-dimensional causal inference and counterfactual prediction. In particular, we will use representation learning approaches to estimate individual treatment effects and counterfactual Gaussian process approaches to simulate alternative drug target response trajectories for individual patients.