This peer-learning group has the form of a journal club in which we discuss scientific articles related to the use of artificial intelligence in medicine or life sciences. The group meets several times per year (currently online) and you can sign up or leave anytime. There is no mandatory attendance and no credits will be given for taking part in this peer-learning group. However, every member is expected to present an article at some point.
Every meeting lasts one hour and focuses on one scientific article, which everyone has read before. The emphasis is on the technical aspects on the paper even though applications will also be discussed to some extent. The person presenting that week gives some background to the paper and summarizes the most interesting technical aspects, with free discussions by the group in between. Papers are mailed out the week before the presentation and pre-approved by the group organizers to ensure they meet the following journal club criteria:
- The article can be understood by people who work with machine learning in medicine/life science, even if they are not experts in the specific area of the article.
- The article presents one or several AI tools with an application in medicine/life science (it is not necessary that the articles concerns a medical problem but it should clearly be possible to apply the tools presented in the article to a medical problem).
- The overall quality of the paper is high (even though there may be some flaws).
- Models are evaluated appropriately, e.g. by comparing with other tools that were designed for the same task.
- The study is conceptually interesting/novel.
- Open source code related to the article is available.
-
25th Oct, 2022 Tuesday @ 13:00
Paper: Medical Coding with Biomedical Transformer Ensembles and Zero/Few-shot Learning (ACL 2022)
-
15th Dec, 2022 Thursday @ 13:00
Paper: Pan-cancer image-based detection of clinically actionable genetic alterations (Nature Cancer)
Link: https://www.nature.com/articles/s43018-020-0087-6#Sec15
Code: https://github.com/jnkather/DeepHistology/releases/tag/v0.2
-
12th Jan, 2023 Thursday @ 13:00
Paper: Pathway importance by graph convolutional network and Shapley additive explanations in gene expression phenotype of diffuse large B-cell lymphoma
-
9th Feb, 2023 Thursday @ 13:00 [Cancelled]
-
9th Mar, 2023 Thursday @ 13:00
-
13th Apr, 2023 Thursday @ 13:00
-
11th May, 2023 Thursday @ 13:00
-
8th Jun, 2023 Thursday @ 13:00
zoom links will be sent out together with the paper to people who have joined this peer-learning group.
-
27th Jan, 2022 Thursday @ 13:00
Paper: Multi-domain translation between single-cell imaging and sequencing data using autoencoders
-
17th Feb, 2022 Thursday @ 13:00
Paper: Instance-based Vision Transformer for Subtyping of Papillary Renal Cell Carcinoma in Histopathological Image
Link: https://arxiv.org/abs/2106.12265
Code: https://github.com/ZeyuGaoAi/Instance_based_Vision_Transformer
-
17th Mar, 2022 Thursday @ 13:00
Paper: CODEX, a neural network approach to explore signaling dynamics landscapes
-
7th Apr, 2022 Thursday @ 13:00
Paper: A deep learning model to predict RNA-Seq expression of tumours from whole slide images
Link: https://www.nature.com/articles/s41467-020-17678-4.pdf
-
24th May, 2022 Tuesday @ 13:00
Paper: Resolving the origins of ancient Eurasian genomes
Link: Only available as an unpublished (submitted) manuscript. Contact Rafsan for a copy.
Code: https://github.com/sarabehnamian/Origins-of-Ancient-Eurasian-Genomes
-
16th Jun, 2022 Thursday @ 13.00
Paper: Deep learning for predicting respiratory rate from biosignals
Code: https://github.com/sydney-machine-learning/biosignal_deeplearning
zoom links will be sent out together with the paper to people who have joined this peer-learning group.
It is possible to join the peer-learning group anytime through the following google form: https://forms.gle/Qw6eatsF1YG2BoC36
You can also register by emailing your full name, positiion and affiliation to the group organizers Rafsan Ahmed (rafsan.ahmed [at] med.lu.se) and Salma Kazemi Rashed (salma.kazemi_rashed [at] med.lu.se).
To participate in the peer-learning group, you also need to register as a COMPUTE member. To do this, simply submit the sign-up form according to the instructions on the COMPUTE website: http://www.compute.lu.se/
You will be signed up automatically as a COMPUTE associate.
After your registration, we will add you to the "COMPUTE members" team on Microsoft Teams. After connecting, please click on the channel "AI4MedLife journal club" and introduce yourself. Use the Teams channel to discuss with your peers in between meetings. You can also share other relavant info, e.g. links to interesting papers you find, your notebooks with coding projects or additional training material.
Instructions for how to use Microsoft Teams can be found here: https://www.staff.lu.se/organisation-and-governance/coronaviruscovid-19-information-regarding-staffphd-doctoral-students/working-home-tools-and-tips/quick-guide-microsoft-teams
The group organizers will get in touch with further instructions. You are also expected to sign up for an article presentation yourself in this sheet: https://github.com/COMPUTE-LU/PLGroup_AI4MedLife-journalclub/blob/main/presentationlist.txt. You can add your name to the fall presentations if you prefer to listen to a few journal clubs first.
https://github.com/Aitslab/training
https://github.com/COMPUTE-LU/AI4MedLife_imaging_2021
COMPUTE is a research school with a focus on scientific discovery using computing, but in the widest sense, meaning any research using digital tools or e-Science for short. COMPUTE brings together partners from several different departments at Lund University. It organizes PhD courses, seminars, workshops and other activities.
Sonja Aits (COMPUTE study director)
Rafsan Ahmed (PhD student)
Salma Kazemi Rashed (PhD student)