I am interested to help. let me know what I can do @dr.richard.barker
1 Like
I would love to help! What can I do?
1 Like
Hello everyone thanks for your interest in collaborating on this meta-analysis.
I’ve created this project management gitbook to help us communicate and share data / results / ideas. This gitbook contains a page "welcome to the team" which contains links to some recommended training for tools that will help you access the code & data shared in the github repo that the book is linked to. You can “fork” this repo into your own github account allowing us to all work synergistically. The book also contains a link to a project planning document, please add your contact details, goal for this project and any notes that come to mind.
There is also a spreadsheet so important ideas/goals can be converted into workpackages for subject matter experts. People are encouraged to volunteer for a role in this new “Microgravity analogue review focus group”, if we have a “communications chair” volunteer we poll when is good for the group to meet and discuss next steps.
1 Like
This is great work…
It looks like we have a few volunteers to help.
Please add your notes and goals to the project management system i just added. Then hopefully we can arrange to get the team that’s forming in this thread to meet up. I hope you’re then willing to explain you progress so far so we can develop some group work packages/ goals to accerlate your progress to help produce some new insights and related academic products. 
1 Like
Ok, hold on, I am developing a system to manage the papers, you will be able to provide access for human labeling (all processes are reversible in case of errors), I will run parallel processes, all human activity will not affects the activity of the machine and vice versa, even if someone in AI/ML wants to provide a method it could be running in parallel and humans can check the labeling of the machine, to get more productivity the system will prioritize unlabeled papers.
The status right now is that I’m going to upload the models I used (but with a better structure to be read because I changed a lot of things yesterday) and I’ll send you the admin interface and instructions for the system when I’m done, I think in a couple of hours, this system is closed because this forum is open, this is to take less risk.
1 Like
Sorry… a lot of meetings. I’m testing and improving the system, I will send you the information at night when I’m back.
1 Like
Added my name to the sheet. Happy to help with this effort!
1 Like
How are these false positives being determined? Manually? Maybe I am just missing context.
1 Like
I’m not sure, so I’m interested to learn about @angel ’s approach… To help I provided a list of about 170 that I “manually checked”. Posted into the GitHub repo and a link earlier in this thread. To be honest I’m surprised the original code/search pulled so many false positives, hence posting the code into the repo incase adding more terms in the original search proves be more accurate or efficient?
Are you trying to gather papers that are spaceflight-adjacent only? There are bound to be a ton of papers using MMG to optimize cell culture in non-spaceflight contexts.
How are you deciding which papers are false positives? For example, this paper from your list of false positives is obviously spaceflight-related, but was it excluded because the RWV mention is actually a reference to a previous paper, and not the primary source for RWV data?
1 Like
Thanks, that’s a great catch. It’s so easy to mistake a paper when one is looking through a list of 17411 paper, that’s why the manual approach was deemed risky.
I wasn’t looking for errors, I am genuinely curious! Just trying to orient to the goal of the effort so I can see how I can contribute.
1 Like
I can take a look Monday. But maybe an approach would be to create a list of keywords that identify applications of interest (spaceflight, space+flight, space+mission, etc) and use Python to filter through your CSV and remove/write those rows into a new file? Process of elimination. Then manually sort through those that are left. Something like this?
2 Likes
I would love to be of assistance. Can I lend a hand with this challenge?
2 Likes
I’m here, I’m going to share two methods (I constantly improve the methods, that’s why I hadn’t uploaded them, in fact I will continue to improve them along the way) that take adventage of the new labeled content:
M1 (faster):
https://www.grupoalianzaempresarial.biz/nasa/awg/microgravityanalogueliteraturereview/m1.txt
M2 (slower):
https://www.grupoalianzaempresarial.biz/nasa/awg/microgravityanalogueliteraturereview/m2.txt
The system is ready. You can ask for access to @dr.richard.barker
2 Likes
@botanynerd is there a list of pubs in a csv from this paper that you can share? (Sorry if there is one referenced in the paper, I do not have access to the pub)
I wonder if we can use the machine learning model that @angel developed for identifying false positives, then use the known list of microgravity analogues from the above pub to train a model for known papers, and cross reference the lists for an extra layer of checking before doing more manual work.
I added the review pdf to the repo 
Chapters 16 and 18 are really good. I extracted the references and saved them in the project spread sheet und er the Zhang et al., 2022 tab, i think/hope they make a great list of positive control papers.
3 Likes
Sorry for the delay in my response, I can use your publication to label relevant papers, even if the references are not listed in NIH, I will develop some tools to extract this specific information.
1 Like
Yes, I’m going to develop another process to integrate the papers in the publication as relevant labeled, I run the process in the background, this does not affect human labeling, the human labeling is really important to obtain a better result, the automated processes could be recalibrated on the way.
1 Like