Back to News
Data Challenge

ADDI NeuroToolKit Data Hackathon

Investigating biomarkers in Alzheimer’s disease

The ADDI NeuroToolKit (NTK) Data Hackathon was a virtual event, bringing together researchers, biostatisticians, data scientists, and clinicians to investigate the potential clinical utility of different biomarkers in Alzheimer’s disease.

Self-selected teams used a beta version of the NTKApp, EPAD datasets, and statistical analysis tools available on the AD Workbench. Participants also accessed our new online community, to engage with team members and other participants, and access Hackathon information and resources.

Specifically, teams:

  • Performed exploratory data analysis
  • Executed standardized descriptive analysis
  • Conducted standardized hypothesis-related analysis
  • Created their own standardized analysis
  • Shared and validated the customization analysis in different datasets

During the Hackathon, teams used specified tools, resources, and datasets to complete the first challenge. The second challenge was optional.

CHALLENGE 1 : ATN Framework (Required)

Biomarkers have become an essential component of Alzheimer’s research.

The ATN (amyloid, tau, neurodegeneration) framework is a descriptive system to categorizing multidomain biomarker findings at the individual person level in a format that is easy to understand. It calls for the seven major Alzheimer’s disease biomarkers to be divided into three binary categories based on the nature of the pathophysiology that each measure.

  • “A” is for beta-amyloid (CSF or PET)
  • “T” is for tau (CSF pTau or tau PET)
  • “N” is for neurodegeneration (CSF tTau, FDG-PET, structural MRI, and others)

Each biomarker category is marked as positive or negative and can be used to profile the etiology of the disease.

The EPAD dataset contains a vast range of data on cognitive outcomes, including socio-demographic data, cognitive assessment results, neurological diagnostic testing and imaging, psychological assessments, medical history and physical examinations, stress, sleep, and quality of life, and life events. These outcomes are hypothesized to be associated with the earliest stages of neurodegenerative disease.

The required ATN framework challenge was for teams to:

  • Explore how the ATN criteria are associated with profiles that may include cognitive, functional, and neuropsychiatric features in this high-risk population. An additional approach is to explore the converse associations of cognitive, functional, and neuropsychiatric features with ATN criteria (or other cognitive factors) may provide additional insights into how patients may be clustered into different risk profiles.
  • Evaluate the different biomarker definitions of A, T, and/or N and propose an improved definition of the ATN factors and compare them to the ATN definitions in published studies (see, “Application of the ATN classification scheme in a population without dementia: Findings from the EPAD cohort,” by Silvia Ingala et. al., Alzheimer’s & Dementia, July 2021 or for example, “ATN classification and clinical progression in subjective cognitive decline,” by Jarith Ebenau, et. al., Neurology, July 7, 2020.)

CHALLENGE 2 : Additional Diagnostics and Classifications (Optional)

Pathophysiological changes usually occur decades before a person experiences Alzheimer’s symptoms and are associated with several modifiable risk factors (see “Dementia prevention, intervention, and care,” by Gill Livingston, et. al., The Lancet, Dec. 16, 2017).

As pathophysiological changes precede cognitive symptoms, interventions (medication or lifestyle changes) can be the most effective when we can predict the disease’s progression.

Our optional diagnostics challenge is for teams to explore the relationship between biomarkers and additional Alzheimer’s disease risk factors, and how this relates to the progression of Alzheimer’s disease. More specifically, teams will:

  • Identify factors or variables that best predict that best predict biomarker (BM) amyloid positive from BM amyloid negative in older, cognitively normal (CN) adults.
  • Define different cutoffs as compared to Elecsys cutoffs.
  • Find individual or combinations of factors or variables that best explains the cognitive performance in older, CN adults.
  • Identify different ways to classify older, CN adults based on risk factors for cognitive impairment.
  • Evaluate the potential clinical use of Abeta42, pTau, and tTau in older, CN adults.

Teams used the new NTKApp, along with either RStudio or Jupyter Notebooks.

RStudio and Jupyter Notebooks are currently available on the AD Workbench. 

NTKApp

A beta version of the NTKApp allows users to curate, analyze, and compare biomarker datasets. The app includes these modules:

  • NTK Curate, enabling data to be curated before being uploaded to a workspac,
  • NTK Analysis, facilitating statistical analysis and provides visualization options
  • NTK Meta Analysis, allowing for data comparisons

RStudio

RStudio is an advanced version of R programming. It is considered an Integrated Development Environment (IDE) that provides a one-stop solution for all statistical computing, graphics, and R scripting in a single interface.

Jupyter Notebooks

A Jupyter Notebook allows users to create and share documents that integrates live code, equations, computational outputs, visualizations, and other multimedia resources, along with explanatory text in a single document. It can be used with Python, R, and other programming languages.

Alzheimer’s is a global disease; it will require the efforts of a global research community to make meaningful progress in finding new diagnostics, therapies, and cures. While spaces in this Hackathon are limited, we encourage researchers from around the world to participate – especially AD Workbench users or NTK Consortium members.

Participants worked in a self-selected team of two to four members and at least one member was experienced in:

  • Programming languages (including R or Python)
  • Biostatistics
  • Neuroscience or Neurology (or a related clinical background that includes an understanding of Alzheimer's disease)
  • Optional: data science (including artificial intelligence or machine learning)

Submissions

After completing the ATN challenge (and optionally the diagnostics challenge), teams produced a short (3-5 minute) video. Each team spokesperson will present their methodology, analysis, and findings.

Judges conducted a blind review (voice and likeness masked) of the videos and score each submission on a scale from 0 (lowest) to 5 (highest) in each of these categories:

  • Patient Value and Clinical Impact
  • Scientific Value
  • Innovation
  • Technical

Judges

This panel of esteemed judges evaluated and scored the Hackathon submissions:

  • Kaj Blennow, Ph.D. Professor and Chief Physician, Department of Psychiatry and Neurochemistry | University of Gothenburg
  • Juan Domingo Gispert, Ph.D. Group Leader, Alzheimer’s Prevention Program – Neuroimaging Research Group | BarcelonaBeta Research Center
  • Sterling Johnson, Ph.D. Professor, Department of Medicine | University of Wisconsin-Madison
  • Craig Ritchie, Ph.D. Professor and Chair of the Psychiatry of Ageing and Director of the Centre for Dementia Prevention | The University of Edinburgh
  • Ingrid van Maurik, Ph.D. Senior Research Associate, Epidemiology and Data Science | Amsterdam UMC – University Medical Centers
  • Lisa Vermunt, Senior Research Associate, Neurochemistry laboratory, Amsterdam Neuroscience | Amsterdam UMC – University Medical Center

Community Contributions

The world cannot bring an end to Alzheimer’s disease if researchers work alone. The “Community Team Contribution” category aimed to recognize teams that break down silos during this Hackathon.

In addition to the blind review, each team received a score based on the value of their contributions to the new community platform. This included the quality and number of posts, and the engagement received from other participants during the Hackathon.

While this Hackathon provided an exciting opportunity for participants to build their analytical skills and contribute to Alzheimer's research, it was also to have a fun reward to work towards. Based on their final scores, teams were awarded the following prizes (a single team cannot win more than one prize):

First and second place teams in the Patient Value and Clinical Impact and Scientific Value category received a Sony Playstation 5

  • Patient Value and Clinical Impact
    First Place: Team Wicking (Australia)
    Second Place: Team UW Badgers (United States)
  • Scientific Value
    First Place: Team Barcelona Neuro Hacker(Spain)
    Second Place: Team TWN (United States)

First and second place teams in the Innovation category received an X-Box Series X

  • First Place: Team NEUROPHET (South Korea)
    Second Place: Team CSIRO (Australia)

First and second place teams in the Technical category received a DJI Mavic Mini 2

  • First Place: Team MD Eagles (United States)
    Second Place: Team Tech Gnomes (United States)

First and second place teams in the Community Team Contribution category received an Oculus Quest 2

  • First Place: Team Laboyama (United States)
    Second Place: Team Team Neuro Machine (India)

Team Wicking and Team MD Eagle's video can be viewed here.