Hackathon
W3PHIAI-23 Aging Hackathon
Background
The identification of biomarkers to describe, define and predict biological age is a very active topic of research. Aging affects organisms differently, and chronological age does not always coincide with biological age, a statement all the more true for diseases like progeroid syndromes and other accelerated aging conditions. Having reliable predictors of chronological age, biological age, and their relationship is important for both diagnostic and prognostic uses (e.g., for disease co-morbidity and mortality), as well as research and clinical use (e.g., rejuvenation and reprogramming, therapeutic outcome, etc.)
Current age indicators rely on molecular and genetic data, usually for biological age, like methylation data (Horvath, Steve, and Kenneth Raj. "DNA methylation-based biomarkers and the epigenetic clock theory of ageing." Nature Reviews Genetics 19.6 (2018): 371-384.) and telomere length (Marioni, Riccardo E., et al. "The epigenetic clock and telomere length are independently associated with chronological age and mortality." International journal of epidemiology 45.2 (2016): 424-432.), mobile phone data data (Felbo, Bjarke, et al. "Using deep learning to predict demographics from mobile phone metadata." (2016).), X-ray scans (Al Zoubi, Obada, et al. "Predicting age from brain EEG signals—A machine learning approach." Frontiers in aging neuroscience 10 (2018): 184.) and face images (Tiwari, Rajesh Kumar. "Human age estimation Using Machine Learning Techniques." International Journal of Electronics Engineering and Applications 8.1 (2020): 01-09.) for chronological age.
Objectives
This hackathon aims at challenging the machine learning community to design an optimal age predictor. There is no limitation or restriction on the type of data used for this challenge, and participants are encouraged to explore the use of multimodal data sources and methods to achieve the goal.
Ground rules
Each team will need to provide an AI-based analysis pipeline capable of characterizing human chronological or biological age. Each group will be asked to select an evaluation strategy of their choosing (e.g., leave-one-out, leave-two-out, k-fold cross validation, etc.).
Process
Teams will have to pre-register by 01/05/2023 to the hackathon via the following form (https://forms.gle/81PyPuw5Jz2Y7Gw97). Teams will need to provide team name, composition and title of their work upon registration. An abstract describing the work will need to be submitted by 01/27/2023.
Workshop
Every team submitting an abstract will be asked to present their work and results at the workshop in a dedicated session (modalities TBD).
Post-Workshop
A selected number of contributions will be considered for inclusion on a special issue of the Experimental Biology and Medicine journal.
Evaluation Criteria
The work of each team will be evaluated by a panel composed by the co-organizers and a selected number of external experts, which will assess it based on originality, reproducibility and performance metrics as presented by the authors.
Prizes
Prizes will be awarded based on originality, reproducibility and performance metrics to the best 3 contributions. The first prize will be 1000$, the second 500$, and the third 250$.
*** 1st Prize ***
Estimating the Brain’s Inherent Age From Sleep-Based EEG Scans
Srikar Katta, Harsh Parikh, Alex Volfovsky, Cynthia Rudin, Haoqi Sun, Brandon Westover
*** 2nd Prize ***
Challenges Facing the Explainability of Age Prediction Models: Case Study for Two Modalities
Mikołaj Spytek, Weronika Hryniewska-Guzik, Jarosław Żygierewicz, Jacek Rogala, Przemyslaw Biecek
*** 3rd Prize ***
Ensuring Age Clock Models are not Attending to Erroneous Features for Improved Biological Age Prediction
Maha M. Alwuthaynani, Amarpal Sahota, Enrico Werner, Jeffrey N. Clark
Background
The identification of biomarkers to describe, define and predict biological age is a very active topic of research. Aging affects organisms differently, and chronological age does not always coincide with biological age, a statement all the more true for diseases like progeroid syndromes and other accelerated aging conditions. Having reliable predictors of chronological age, biological age, and their relationship is important for both diagnostic and prognostic uses (e.g., for disease co-morbidity and mortality), as well as research and clinical use (e.g., rejuvenation and reprogramming, therapeutic outcome, etc.)
Current age indicators rely on molecular and genetic data, usually for biological age, like methylation data (Horvath, Steve, and Kenneth Raj. "DNA methylation-based biomarkers and the epigenetic clock theory of ageing." Nature Reviews Genetics 19.6 (2018): 371-384.) and telomere length (Marioni, Riccardo E., et al. "The epigenetic clock and telomere length are independently associated with chronological age and mortality." International journal of epidemiology 45.2 (2016): 424-432.), mobile phone data data (Felbo, Bjarke, et al. "Using deep learning to predict demographics from mobile phone metadata." (2016).), X-ray scans (Al Zoubi, Obada, et al. "Predicting age from brain EEG signals—A machine learning approach." Frontiers in aging neuroscience 10 (2018): 184.) and face images (Tiwari, Rajesh Kumar. "Human age estimation Using Machine Learning Techniques." International Journal of Electronics Engineering and Applications 8.1 (2020): 01-09.) for chronological age.
Objectives
This hackathon aims at challenging the machine learning community to design an optimal age predictor. There is no limitation or restriction on the type of data used for this challenge, and participants are encouraged to explore the use of multimodal data sources and methods to achieve the goal.
Ground rules
Each team will need to provide an AI-based analysis pipeline capable of characterizing human chronological or biological age. Each group will be asked to select an evaluation strategy of their choosing (e.g., leave-one-out, leave-two-out, k-fold cross validation, etc.).
Process
Teams will have to pre-register by 01/05/2023 to the hackathon via the following form (https://forms.gle/81PyPuw5Jz2Y7Gw97). Teams will need to provide team name, composition and title of their work upon registration. An abstract describing the work will need to be submitted by 01/27/2023.
Workshop
Every team submitting an abstract will be asked to present their work and results at the workshop in a dedicated session (modalities TBD).
Post-Workshop
A selected number of contributions will be considered for inclusion on a special issue of the Experimental Biology and Medicine journal.
Evaluation Criteria
The work of each team will be evaluated by a panel composed by the co-organizers and a selected number of external experts, which will assess it based on originality, reproducibility and performance metrics as presented by the authors.
Prizes
Prizes will be awarded based on originality, reproducibility and performance metrics to the best 3 contributions. The first prize will be 1000$, the second 500$, and the third 250$.
*** 1st Prize ***
Estimating the Brain’s Inherent Age From Sleep-Based EEG Scans
Srikar Katta, Harsh Parikh, Alex Volfovsky, Cynthia Rudin, Haoqi Sun, Brandon Westover
*** 2nd Prize ***
Challenges Facing the Explainability of Age Prediction Models: Case Study for Two Modalities
Mikołaj Spytek, Weronika Hryniewska-Guzik, Jarosław Żygierewicz, Jacek Rogala, Przemyslaw Biecek
*** 3rd Prize ***
Ensuring Age Clock Models are not Attending to Erroneous Features for Improved Biological Age Prediction
Maha M. Alwuthaynani, Amarpal Sahota, Enrico Werner, Jeffrey N. Clark