News and Updates
To subscribe to announcements and ask questions, sign up on the TADPOLE Google Group
4 Feb 2019: ETA for evaluation coming soon! The TADPOLE team is working with ADNI to assess the ADNI3 rollover data and its readiness for TADPOLE evaluation. We will update you soon. Stay tuned.
11 May 2018: The paper describing the design of TADPOLE Challenge has been published:
TADPOLE Challenge: Prediction of Longitudinal Evolution in Alzheimer's Disease, Razvan V. Marinescu, Neil P. Oxtoby, Alexandra L. Young, Esther E. Bron, Arthur W. Toga, Michael W. Weiner, Frederik Barkhof, Nick C. Fox, Stefan Klein, Daniel C. Alexander, the EuroPOND Consortium, arXiv:1805.03909, 2018
17 November 2017: The TADPOLE submission system remains open, but submissions after 16 November 2017 will be ineligible for prizes.
16 November 2017: TADPOLE submission is now closed! Thank you all for your submissions. We will continue to accept methods descriptions documents for a short period.
26 September 2017: TADPOLE will be present at the PyCon UK conference on Sunday 29 October. We will organise a hackaton where participants will submit further leaderboard entries, improve our scripts or create augumented datasets which will be shared with everyone. Hackaton github repo here.
13 September 2017: TADPOLE Open Webinar 3 will happen via YouTube Live this Thursday 14 September (tomorrow!) at 1400 London time (UTC + 1) - europond.eu/tadpole-webinar. Please join us and have your questions ready: email@example.com or Euro_POND on Twitter or via the TADPOLE Google Group
30 August 2017: We created a TADPOLE Google Group for replacing the mailing list. Please ask all your questions on the Google Group from now on.
4 August 2017: TADPOLE Open Webinar 2 will happen via YouTube Live this Thursday 10 August at 1400 London time (UTC + 1) - europond.eu/static/tadpolewebinar. Please join us and have your questions ready: firstname.lastname@example.org or Euro_POND
12 July 2017: TADPOLE article on AlzForum: www.alzforum.org/news/community-news/tadpole-challenge-seeks-best-predictors-alzheimers
10 July 2017: TADPOLE Open Webinar 1 to happen this Wednesday 12 July at 1400 London time (UTC + 1) - europond.eu/static/tadpolewebinar. Get your questions ready!
4 July 2017: TADPOLE challenge welcomes a third prize sponsor: Alzheimer's Research UK.
16 June 2017: Data sets available on ADNI site (login to ADNI, follow Download -> Study Data -> Test Data -> Data -> "Tadpole Challenge Data")
15 June 2017: Submission system open
What is TADPOLE?
TADPOLE is a challenge to identify which people within an age group at risk of AD will start to show symptoms in the short to medium term (1-5 years). It focusses on rollover individuals in the Alzheimer's disease neuroimaging initiative (ADNI) study. The challenge is to use historical measurements from these individuals to forecast future measurements. The following scientific questions motivate the challenge:
- How predictable is progression to Alzheimer’s Disease (AD) in at-risk individuals?
- Which data, processing pipelines, and predictive models best predict future AD progression?
- Can we use such methods to improve cohort selection for clinical trials?
We invite you to participate in TADPOLE and join the international fight against dementia and AD.
Why is it important?
AD, and dementia in general, is a key challenge for 21st century healthcare. The statistics are sobering: 7% of people over 65 and 20% over 80 suffer from dementia of which AD is the most common cause; dementia has higher health and social care costs than cancer, stroke and chronic heart disease combined – projected $1T in 2018 and $2T in 2030; a treatment slowing progression by 50% would reduce annual care costs by about 10%, i.e., $100B in 2018.
No current treatments provably cure or even slow AD. Of the hundreds of clinical trials into putative treatments, often running to billions of dollars/euros, fewer than 1% have proceeded to the regulatory approval stage and none have managed to prove a disease-modifying effect. One key reason for these failures is difficulty in identifying patients at early stages of the disease where treatments are most likely to be effective.
TADPOLE challenges you to identify what data and algorithms best predict AD progression. This facilitates early identification of patients likely to be receptive to treatment, thus purifying cohorts for clinical trials, highlighting positive treatment effects, and facilitating translation of effective treatments to market.
How does it work?
TADPOLE provides a list of individuals at an age that puts them at risk of AD. A history of informative measurements (from imaging, psychology, demographics, genetics, etc.) from each individual is available to inform forecasts. Each individual has agreed to a follow-on assessment.
TADPOLE participants predict future measurements from these individuals and submit their predictions by the submission deadline: 15 November 2017.
Once future measurements are available, TADPOLE will evaluate each forecast against them, collate and compare results, publish the results, award prizes and, ultimately, write up a scientific paper co-authored with active participants that input sufficient information.
We encourage forecasts from any source: statistical, mathematical, or computational models; or simply informed guesses by human experts. The competition is open to anyone: from statisticians to neurologists; from industry or academic research; from professor to high-school student.
Join the challenge!
The challenge was open for submissions from 15 June to 15 November 2017. The submission system remains open (in particular, the Leaderboard is a useful tool for those developing forecasting tools), but entries received after 15 November 2017 are not eligible for prizes, nor for coauthorship on scientific publications emanating from the challenge.
Consultation on the details of the challenge was open from 15 June 2017 until 15 August 2017 after which the rules will be fixed. We welcome thoughts and suggestions on all aspects of the challenge before then – please visit the Google Group, or email email@example.com.