Saameh Sanaaee
LightNews LightNews
banner
saamehsanaaee.bsky.social
Saameh Sanaaee
@saamehsanaaee.bsky.social
210 followers 220 following 190 posts
Neurobiology, Neuroimaging, and Molecular Biology

(she/her)

https://saamehsanaaee.github.io/
Posts Replies Media Videos
saamehsanaaee.bsky.social
Did a thing today!

Notes on Reading Papers

I've compiled these from multiple sources with a dash of experience (as much experience as you can expect from a junior scientist).
Notes on Reading Papers (page 1) Compiled by a junior scientist from various resources. A 2-page document describing the sequence of reading papers. The bottom line up front: 1. Read Title and Authors' names to get an understanding of different labs and the groups that are focusing on different topics and select interesting papers. 2. Read the Abstract to short-list the papers you want to dive into. 3. Study Figures and their captions to get a general understanding of the work. 4. Depending on your expertise in the field, move on and skim the Results or go through the Conclusion or Introduction 5. Skim the Discussion and Methods. By the end, you should understand the hypothesis, methods, findings, and limitations. Continuation: Page 2 of the Notes on Reading Papers document. Compiled by a junior scientist from various resources. The bottom line up front: 1. Read Title and Authors' names to get an understanding of different labs and the groups that are focusing on different topics and select interesting papers. 2. Read the Abstract to short-list the papers you want to dive into. 3. Study Figures and their captions to get a general understanding of the work. 4. Depending on your expertise in the field, move on and skim the Results or go through the Conclusion or Introduction 5. Skim the Discussion and Methods. By the end, you should understand the hypothesis, methods, findings, and limitations.
November 10, 2025 at 8:00 PM
saamehsanaaee.bsky.social
This post was the final straw for me!

Although, the entity reading this text is probably going to be an "AI agent" which has transformed many full-time jobs into contracts with not regard for humanity.
Deleting Academia account with this text as my reason: I believe Academia's decisions on AI-based tools or promotions or whatever the corporate term is for "destroying proper science communication" has made it unbearable to use this platform. I hope you come to your senses, but by all means, continue destroying people's work until then like it's not a giant mistake.
November 1, 2025 at 11:16 AM
saamehsanaaee.bsky.social
"I've got 69 cites."

"You know what that means."

"I don't have enough cites for ... wait, I actually have 70!"
Screenshot of my Google Scholar header.
October 15, 2025 at 12:11 PM
saamehsanaaee.bsky.social
New favorite cat photo that I found when looking up "Electrostatic Potential Energy":
A cat with a determined face that could say, "And I'll do it again" with Styrofoam peanuts stuck all over its body. And here's the alt text copied directly from Wikipedia: "Styrofoam peanuts clinging to a cat's fur due to static electricity. The triboelectric effect causes an electrostatic charge to build up on the fur due to the cat's motions. The electric field of the charge causes polarization of the molecules of the styrofoam due to electrostatic induction, resulting in a slight attraction of the light plastic pieces to the charged fur." Original image: Sean McGrath from Saint John, NB, Canada
July 13, 2025 at 12:16 PM
saamehsanaaee.bsky.social
Bliss!
A screenshot from Google Play Store showing Zotero, the research assistant and organizer, installed on my phone.
June 29, 2025 at 12:54 PM
saamehsanaaee.bsky.social
Is it just me or has the frequency of fires (that contribute to "the world is on fire") increased around the world?
From KC Green's web comic strip "On Fire," became a popular online meme by 2014
June 13, 2025 at 6:54 AM
saamehsanaaee.bsky.social
Today, I got my diploma.

I graduated on February 4, 2024.

489 days.

If europeans complain about bureaucracy in front of me, I'll punch a wall.
June 7, 2025 at 6:51 PM
saamehsanaaee.bsky.social
Here's what I'll say:
- The course registration that has been auto-magically withdrawn is from CDC.
- Train(dot)org is CDC's training and certification platform.
And, finally, I'm not from the US.

I wonder if these facts relate to any of the recent events we've been hearing about!
May 30, 2025 at 8:44 AM
saamehsanaaee.bsky.social
Our main figure:
The main figure of the publication that shows the sequence of our models and their outputs. Here's the figure legend from the paper: Figure 1. From parcel-based BOLD signals to predictions A. WMD Model: A-1. An MLP with a funnel-shaped structure takes in average parcel-based BOLD signals. A-2. The WMD achieved an accuracy of 90% . B. WMAD Model: B-1. Subject-based adjacency matrices, parcel-based BOLD timeseries, and target matrices are input into the GAT and the LSTM. Outputs are unified through the fusion layer and passed to an MLP to generate the final parcel-based demand probability. B-2. A 100% accuracy for WMAD, possibly due to overfitting or data sparsity. C. WMD and WMAD results: C-1. WMD predicts math as high-demand, fear and story as low-demand, and does not predict neutral subtask in either category. C-2. WMAD shows similar predictions for math, story, and neutral subtask, but predicts fear as high-demand (C-2, top left).
May 21, 2025 at 7:58 AM
saamehsanaaee.bsky.social
Did I just spend over 30 minutes aligning a dictionary that, probably, only I'll be looking into?

Yes, I did. Why?
Because as Marie Kondo said ...

This one sparks joy. | This one doesn't.
A dictionary of the Glasser parcellation in python aligned to have the same space for each entry. "This one sparks joy." The same dictionary of the Glasser parcellation in python not aligned and has lines with varying length and number of entries. "This one does not spark joy."
April 19, 2025 at 9:55 AM
saamehsanaaee.bsky.social
I can't believe that after over TWO CENTURIES, scientists still need to explain to *some* people why vaccines are needed or why, if they actually have any hint of a positive feeling for their children, they need to get them vaccinated.

My point is:
Thank you, Angela Collier, for this video!
Screenshot of the "Timeline of human vaccines" page on Wikipedia: "This is a timeline of the development of prophylactic human vaccines. Early vaccines may be listed by the first year of development or testing, but later entries usually show the year the vaccine finished trials and became available on the market. Although vaccines exist for the diseases listed below, only smallpox has been eliminated worldwide. The other vaccine-preventable illnesses continue to cause millions of deaths each year. Currently, polio and measles are the targets of active worldwide eradication campaigns. 1796 – Edward Jenner develops and documents first vaccine for smallpox." This screenshot was put on the post to emphasize how long humanity has been vaccinated.
March 30, 2025 at 4:39 AM
saamehsanaaee.bsky.social
Not only I included this comment in my git commit ... but I noticed that my teddy bear will be eternally visible in a meeting screenshot that we used as our group photo.

Perfect ... just perfect.

Am I science-ing? Am I legit now?
Screenshot of my git commit message that reads: "Also ... 100th commit! Yay! 🥳 Do we celebrate those? 👀"
March 28, 2025 at 1:03 PM
saamehsanaaee.bsky.social
(4/12)
So, let's not be too critical now, right? Let's see one of the answers.

This question and its answer are very technical. They are clearly not directed towards the layman just to promote science and #ScienceCommunication.
One of the questions: "How do the Homotopic Markov Random Field (hMRF) parcellations compare to other homotopic parcellations in terms of Resting State Functional Connectivity (RSFC)? The Homotopic Markov Random Field (hMRF) parcellations exhibited stronger homotopic Resting State Functional Connectivity (RSFC) than other homotopic parcellations, indicating their enhanced capability in capturing functional connectivity patterns during resting states." This is technical, right? It's too technical for the layman.
March 11, 2025 at 8:29 AM
saamehsanaaee.bsky.social
(3/12)
You know, the abstract; the part of a paper that summarizes what "questions [are] answered in this article" and has been a part of scientific publications for years and years.

If you look closely at the results from the LLM, you can see how technical it is.

Here's the screenshot again:
March 11, 2025 at 8:29 AM
saamehsanaaee.bsky.social
(2/12)
So, if you click on that and scroll past THE ABSTRACT, you see this:
Another screenshot of the actual "questions answered in this article" section. This is shown after the abstract of the paper.
March 11, 2025 at 8:29 AM
saamehsanaaee.bsky.social
Possibly, I'm too furious with the overtake of LLMs, especially when it's just a waste of everyone's time.

This is a thread! 🧵(1/12)

I was looking through to find a paper today and I see this, a section called "questions answered in this article" in beta that is "Powered by GenAI" on Elsevier:
A screenshot of the outline for a paper that shows a section, "questions answered in this article" which is "Powered by GenAI" as a beta functionality.
March 11, 2025 at 8:29 AM
saamehsanaaee.bsky.social
"So far, Schlicht’s YesNoError team has quantified the false positives in only around 100 mathematical errors that the AI found in an initial batch of 10,000 papers."

"Only 100 false positives" is basically "destroying the results from hundreds of hours a team of people have spent on the research."
March 8, 2025 at 6:55 AM
saamehsanaaee.bsky.social
"The rate of false positives, instances when the AI claims an error where there is none, is a major hurdle."

"Each alleged error must be checked with experts in the subject, and finding them is the project’s greatest bottleneck."

This part is self-explanatory.
March 8, 2025 at 6:55 AM
saamehsanaaee.bsky.social
"The cost of analysing each paper ranges from 15 cents to a few dollars, depending on the length of the paper and the series of prompts used."

It's that "Oh, wait, no. You have to pay now." I just mentioned.
March 8, 2025 at 6:55 AM
saamehsanaaee.bsky.social
'"If you start pointing fingers at people and then it turns out that there was no mistake, there might be reputational damage," says Michèle Nuijten, a researcher in metascience at Tilburg University in the Netherlands.'

Again! It's double the work! Not saving time, it's creating more problems.
March 8, 2025 at 6:55 AM
saamehsanaaee.bsky.social
Short and simple:
"Here, use this tool for free!"
"Oh, wait, no. You have to pay now."
March 8, 2025 at 6:55 AM
saamehsanaaee.bsky.social
"... flaws--many of which have yet to be verified by a human ... YesNoError has a plan to eventually do so at scale."

It's double the work!
If peer reviewers are compensated well, they will have enough time and incentive to be thorough and they WILL catch that error BEFORE THEY LET IT BE PUBLISHED.
March 8, 2025 at 6:55 AM
saamehsanaaee.bsky.social
This is how I feel today with all the deadlines and lab tests and checkups.
Picture of a hamster with bedding stuck on its head. To me, it looks like that "my chill is fake" meme. I also got it from Pinterest and I don't know the proper source. Sorry.
March 3, 2025 at 9:44 AM