2024 02.02
【競賽】INTERSPEECH'24 TAUKADIAL Challenge: Speech-Based Cognitive Assessment in Chinese and English
We are happy to announce the INTERSPEECH'24 TAUKADIAL Challenge:
Speech-Based Cognitive Assessment in Chinese and English. The challenge
focuses on creating models for predicting cognitive test scores and
diagnosing mild cognitive impairment (MCI) in elderly speakers of
Chinese and English using samples of connected speech. In addition to
participation in the challenge, we invite the submission of papers on
system ideas (machine learning architectures), novel signal processing
features, feature selection, and feature extraction methods for the
TAUKADIAL Challenge.

Cognitive problems, such as memory loss, speech and language impairment,
and reasoning difficulties, often occur among older adults, preceding
the onset of dementia syndromes. Due to the global prevalence of
dementia, research into cognitive impairment for dementia prevention and
early detection has become a healthcare priority. There is a need for
cost-effective and scalable methods for assessing cognition and
detecting impairment, ranging from subtle forms to severe manifestations
of dementia.  Speech, as an easily collectable behavioral signal
reflecting cognitive function, has the potential to serve as a digital
biomarker of cognitive function, presenting a unique opportunity for the
application of speech technology. While most studies have focused on
English speech data, the TAUKADIAL Challenge aims to explore speech as a
marker of cognition globally, providing data in two major languages:
Chinese and English.

This INTERSPEECH Challenge aims to bring together members of the speech,
signal processing, machine learning, natural language processing, and
biomedical research communities, allowing them to test existing methods
or develop novel approaches on a new standardized dataset to be provided
by us, and made available for future research and result replication.

How to Participate:
First, register for the TAUKADIAL Challenge to gain access to the
TAUKADIAL dataset by emailing with your contact
information and affiliation. Full dataset access will be provided
through DementiaBank membership. To become a member, include in your
email to a general statement of how you plan to
use the data, with specific mention of the TAUKADIAL Challenge. If you
are a student, ask your supervisor to join DementiaBank as well.

The TAUKADIAL Challenge encompasses the following tasks:

1) A classification task: Participants will create models to distinguish
   healthy control speech from MCI speech.

2) A cognitive test score prediction (regression) task: Create a model
   to infer the subject's Mini Mental Status Examination (MMSE) scores
   based on spontaneous speech.

You may choose to do one or both of these tasks. You will receive access
to a training set, and two weeks before the paper submission deadline,
you will receive test sets to evaluate your models.

Saturnino Luz, Sofia de la Fuente Garcia, Fasih Haider (University of Edinburgh),
Davida Fromm, Brian MacWhinney (Carnegie Mellon University),
Ya-Ning Chang (Miin Wu School of Computing, National Cheng Kung University),
Yi-Chien Liu, Chia-Ju Chou (Neurology Department at Cardinal Tien Hospital)

Important Dates:
- 25th January: TAUKADIAL Challenge announced; Call for Participation Published
- 20th February: registration deadline; email to
  register for the challenge and receive the training set.
- 27th February: deadline for submission of results
- INTERSPEECH Paper Submission Deadline: 2 March 2024
- INTERSPEECH paper update deadline: 9 March 2024
- INTERSPEECH paper acceptance notification: 6 June 2024.
- TAUKADIAL Session at INTERSPEECH: TBD (between 2 and 7 September 2024)
See other important dates on the INTERSPEECH: 2024 website.

For further information, see the TAUKADIAL Challenge website