POIS Research Database Project - ~500 detailed case studies extracted

[mineral]

UPDATE: Revised version of this project has been published: poisdata.org

---
Hi POIS community,

Following up on my previous meta-analysis of POISCenter data, I'm excited to share that I've created a public POIS research database that's now available online:


https://publish.obsidian.md/pois-research/index

The database includes:

• ~500 detailed case studies extracted
• 1,400+ Reddit posts/comments analyzed 
• 4,400+ POISCenter posts/comments processed

Each case study is structured to include:

• Timeline and symptom progression
• Treatment attempts and outcomes
• Identified triggers

The goal is to help both researchers and POIS sufferers better understand patterns in symptoms, treatments, and outcomes.
Thanks to everyone who has shared their experiences - your contributions have made this knowledge base possible.

[demografx]

Thanks, Mineral! Sent your post and previous meta-analysis to our POIS Research Team

[Quantum]

Thanks, Mineral, for your great work ! And thank you for putting it online, as well


Your database will be useful for sure.

Considering the current state of machine learning, as well as AI, those who will consult it should be aware that all data is not extracted to the database, and also that some inaccuracies have been introduced.  This is in no way a negative critic of your work, Mineral - you did a great job !  My intention is that those who will consult your database will be better prepared to interpret and use what they see.

I find it very useful that you have included information and cases from the POIS Reddit.  The format of the Reddit forum makes it hard to dig and have an overview.

[Warrior]

Mate, this is amazing! Thank you

Just want to say one thing tho - on my page (https://publish.obsidian.md/pois-research/case-studies/tteezzkk-481), it lists medications - Zyrtec and NSAIDS provided partial relief. This isn't true, and I've never actually stated this online (I've never tried either of them). Not trying to bring this down at all because I think the work here is amazing and definitely a step in the right direction, but it does make me question if there is a % of error within the data?

[Sisyphus]

Wow, that's brilliant!
It looks like you've put a lot of work into this. I wish I had your IT skills!
That's great you've had positive feedback from forum admin.
I think what mineral has started is fantastic and hopefully it will add an extra dimension to this forum.

[b_jim]

Excellent job.    No doubt it will be useful for our communuty and it could be a reference for all Poisers in the world. A real tool  !

[demografx]

Thanks, Mineral, for your great work ! And thank you for putting it online, as well


Your database will be useful for sure.

Considering the current state of machine learning, as well as AI, those who will consult it should be aware that all data is not extracted to the database, and also that some inaccuracies have been introduced.  This is in no way a negative critic of your work, Mineral - you did a great job !  My intention is that those who will consult your database will be better prepared to interpret and use what they see.

I find it very useful that you have included information and cases from the POIS Reddit.  The format of the Reddit forum makes it hard to dig and have an overview.

I second Quantum’s remarks!!

Thank you, Mineral!

Best,
Demo

[carlo_francesco]

Wow  , thank you for your effort. Since the entries seem to be curated by yourself and not automatically extracted, this must have been quite a lot of work. The tagging functionality of obsidian really lends itself for thematical filtering, allowing for quick overviews of specific topics related to POIS. I will certainly take some time to browse through some topics I am interested in.
One critique point would be that, because of the lacking references, the entries need to be taken at face value. I know that it is quite tedious, but have you considered adding references/links ex-post to the entries?

[b_jim]

Do you think some non-professional like us can publish our own scientific study leaning on mineral's database ?

Some sites like medrxiv accept to publish preprint studies. I tell myself that if we were doing serious work we have a chance to be accepted...

[Sisyphus]

Yes, I like this idea!
I think there is a brilliant opportunity right before our eyes for this forum to do an internal research project involving mineral's db or a further development of it. So many pros.

[demografx]

Do you think some non-professional like us can publish our own scientific study leaning on mineral's database ?

Some sites like medrxiv accept to publish preprint studies. I tell myself that if we were doing serious work we have a chance to be accepted...

https://www.perplexity.ai/search/can-non-professionals-publish-UOBD3nUDR96FWvykANTbBg

[demografx]

Yes, I like this idea!
I think there is a brilliant opportunity right before our eyes for this forum to do an internal research project involving mineral's db or a further development of it. So many pros.

Thanks, Sisyphus!

Please see my reply to b_jim:
https://www.perplexity.ai/search/can-non-professionals-publish-UOBD3nUDR96FWvykANTbBg

[b_jim]

Thank you for this excellent response.
One day the problem will be solved. And it would bother me if there wasn't a trace of our work.

[demografx]

Thank you for this excellent response.
One day the problem will be solved. And it would bother me if there wasn't a trace of our work.

Quantum, the admins, and our computer hosting service are also taking frequent steps to ensure that all our historical forum data is saved and categorized!

For our pre-POISCenter data (@ NSF) we need to work with the owner in the UK.

[mineral]

Thanks, Mineral! Sent your post and previous meta-analysis to our 2025 POIS Scientific Research Team

That's awesome, thanks demografx!

Thanks, Mineral, for your great work ! And thank you for putting it online, as well


Your database will be useful for sure.

Considering the current state of machine learning, as well as AI, those who will consult it should be aware that all data is not extracted to the database, and also that some inaccuracies have been introduced.  This is in no way a negative critic of your work, Mineral - you did a great job !  My intention is that those who will consult your database will be better prepared to interpret and use what they see.

I find it very useful that you have included information and cases from the POIS Reddit.  The format of the Reddit forum makes it hard to dig and have an overview.

Thanks Quantum, I did see many inaccuracies in the data that were introduced by the data pipeline. I've revised the system data flow, improved classification, aggregation and added links to original posts for reference to improve the overall accuracy of the data. The system is still not perfect, you're correct that when interpreting the data one should assume some inaccuracies might still exists, i.e missing or incorrect information.

Mate, this is amazing! Thank you

Just want to say one thing tho - on my page (https://publish.obsidian.md/pois-research/case-studies/tteezzkk-481), it lists medications - Zyrtec and NSAIDS provided partial relief. This isn't true, and I've never actually stated this online (I've never tried either of them). Not trying to bring this down at all because I think the work here is amazing and definitely a step in the right direction, but it does make me question if there is a % of error within the data?

Hi Warrior Thank you for the comment.
Getting an actual error % on the data is quite complex, I did managed to improve the data accuracy by a major factor in the latest version. In your new case study page I can see no mention of Zyrtec and NASAID