Shaoshi Zhang (@ZShaoshi)

Pinned

S

Thrilled to share our latest work just published in @Nature where we looked into the optimal fMRI scan time for brain-wide association studies (BWAS)! Full thread below 👇

TThomas Yeo@bttyeo · Jul 17

1/11 Excited to share our @Naturestudy led by @Leon_Oo1 @csabaorban @ZShaoshi doi.org/10.1038/s41586… It is well-known that AI performance scales with logarithm of sample size (Kaplan, McCandlish 2020), but in many domains, sample size can be # participants or # measurements...

1

7

15

0

2.0K

S

Shaoshi Zhang@ZShaoshi · Jul 23

Amazing work from @bttyeo on leveraging compute scaling laws to improve predictions using neuroimaging data!

TThomas Yeo@bttyeo · Jul 17

1/11 Excited to share our @Naturestudy led by @Leon_Oo1 @csabaorban @ZShaoshi doi.org/10.1038/s41586… It is well-known that AI performance scales with logarithm of sample size (Kaplan, McCandlish 2020), but in many domains, sample size can be # participants or # measurements...

0

4

8

0

1.0K

S

Shaoshi Zhang@ZShaoshi · Jul 22

@ten_photos collaborated with researchers at the National University of Singapore on a recent study published in @Nature on how longer duration fMRI brain scans reduce costs and improve prediction accuracy for AI models. Read more about the study below 👇

TThomas Yeo@bttyeo · Jul 17

1/11 Excited to share our @Naturestudy led by @Leon_Oo1 @csabaorban @ZShaoshi doi.org/10.1038/s41586… It is well-known that AI performance scales with logarithm of sample size (Kaplan, McCandlish 2020), but in many domains, sample size can be # participants or # measurements...

0

4

7

2

1.0K

S

Shaoshi Zhang@ZShaoshi · Jul 17

“This is a gamechanger for the field.” A study co-authored by professor of neurology @ndosenbach with @NUSingapore shows how to optimize brain scans for designing smarter and more cost-effective studies of neurological and psychiatric conditions.

nnature@Nature · Jul 17

Nature research paper: Longer scans boost prediction and cut costs in brain-wide association studies go.nature.com/46fWRFe

0

6

10

1

2.0K

Shaoshi Zhang Retweeted

N

Nico Dosenbach@ndosenbach · Jul 18

I think they’re all consistent. In PFM you’re not trying to make predictions about other brains. You’re characterizing individual brains and individual-specific intervention effects. Measurement accuracy and precision are critical. Sample size not at all.

1

2

7

1

724

Shaoshi Zhang Retweeted

N

Nico Dosenbach@ndosenbach · Jul 18

In BWAS you’re trying to make predictions about unseen brains. Sample size is absolutely critical, similar to GWAS. But it turns out that in real-world scenarios because the FC signal naturally varies with time and is noisy, … it’s more cost efficient to scan longer per person.

1

3

8

1

693

Shaoshi Zhang Retweeted

n

nature@Nature · Jul 17

Nature research paper: Longer scans boost prediction and cut costs in brain-wide association studies go.nature.com/46fWRFe

0

9

30

6

13.0K

S

Shaoshi Zhang@ZShaoshi · Jul 17

For me, this work is a classic @OHBM story: In 2023 I wasn't working with @bttyeo but I overheard him at his poster pointing to some scan time accuracy curves on his poster saying "I don't why they have this particular shape". That kicked off the collab that led to these results.

TThomas Yeo@bttyeo · Jul 17

3/11 ... model. Tom's model explains empirical accuracies well across 76 phenotypes from 9 resting-fMRI & task-fMRI datasets (R2 = 0.89), spanning many scanners, acquisitions, racial groups, disorders & ages. Does this mean that we should collect large datasets & short scans?

0

8

28

1

4.0K

S

Shaoshi Zhang@ZShaoshi · Jul 17

another milestone work in the field👍👍👍

TThomas Yeo@bttyeo · Jul 17

1/11 Excited to share our @Naturestudy led by @Leon_Oo1 @csabaorban @ZShaoshi doi.org/10.1038/s41586… It is well-known that AI performance scales with logarithm of sample size (Kaplan, McCandlish 2020), but in many domains, sample size can be # participants or # measurements...

1

3

8

0

985

Shaoshi Zhang Retweeted

L

Lijun AN | 安丽军@anlijuncn · Jul 15

Can AI reveal the risk and co-pathology of multiple neurodegenerative diseases from just a single blood sample? We explored the AI-based diagnostic power on a massive sample (N=17,170) and high-rank plasma proteomics data. medrxiv.org/content/10.110… #MedSky #neuroskyence #neurosky…

2

15

34

5

5.0K

S

Shaoshi Zhang@ZShaoshi · Jul 17

Super thankful to @bttyeo @csabaorban and @ZShaoshi for pouring in all the effort to make this work possible!

TThomas Yeo@bttyeo · Jul 17

1/11 Excited to share our @Naturestudy led by @Leon_Oo1 @csabaorban @ZShaoshi doi.org/10.1038/s41586… It is well-known that AI performance scales with logarithm of sample size (Kaplan, McCandlish 2020), but in many domains, sample size can be # participants or # measurements...

2

7

14

0

1.0K

S

Shaoshi Zhang@ZShaoshi · Jul 17

Proud to be part of this exciting @Nature study! It's time to embrace longer fMRI scan durations!

TThomas Yeo@bttyeo · Jul 17

1/11 Excited to share our @Naturestudy led by @Leon_Oo1 @csabaorban @ZShaoshi doi.org/10.1038/s41586… It is well-known that AI performance scales with logarithm of sample size (Kaplan, McCandlish 2020), but in many domains, sample size can be # participants or # measurements...

0

10

25

0

3.0K

S

Shaoshi Zhang@ZShaoshi · Jul 17

V useful paper by @bttyeo @Leon_Oo1 & @csabaorban out in @Nature. Scan for longer if you want to predict behaviour using fMRI and save $. Check out their calculator: thomasyeolab.github.io/OptimalScanTim…. Also another great use of our TCP data set (pmc.ncbi.nlm.nih.gov/articles/PMC11…).

TThomas Yeo@bttyeo · Jul 17

1/11 Excited to share our @Naturestudy led by @Leon_Oo1 @csabaorban @ZShaoshi doi.org/10.1038/s41586… It is well-known that AI performance scales with logarithm of sample size (Kaplan, McCandlish 2020), but in many domains, sample size can be # participants or # measurements...

2

8

22

3

1.0K

Shaoshi Zhang Retweeted

M

Marcelo Mattar@marcelomattar · Jul 2

Thrilled to see our TinyRNN paper in @nature! We show how tiny RNNs predict choices of individual subjects accurately while staying fully interpretable. This approach can transform how we model cognitive processes in both healthy and disordered decisions. doi.org/10.1038/s41586…

4

52

261

109

17.0K

S

Shaoshi Zhang@ZShaoshi · Apr 11

Check out our latest preprint led by the amazing @tianchuzeng and @t___fang where we speed up the tedious parameter optimization process for biophysical modelling🔥👇

TThomas Yeo@bttyeo · Apr 11

While the world burns, we cook up a new preprint! doi.org/10.1101/2025.0… Biophysical modeling is a key tool to derive mechanistic insights into the brain. These models are governed by biologically meaningful parameters (unlike deep neural nets), but the dirty secret ... 1/N

0

1

4

0

260

Shaoshi Zhang Retweeted

H

Holmes Lab@HolmesLab_BHI · Feb 19

New paper by Lydia Qu @laurant_lydia et. al out now in Nature @NatMentHealth 📰 Here, we show that predictive network features are distinct across internalizing and externalizing traits/behaviors. With @carrisa_cocuzza,@bttyeo,@elvisha9, and more 🌟 shorturl.at/67TH1

1

21

55

12

4.0K

Shaoshi Zhang Retweeted

T

Thomas Yeo@bttyeo · Nov 20

🚨 Brain Age vs Direct Models in Alzheimer’s disease (AD) 🚨 A thread 🧵 1/ Brain age is a powerful indicator of general brain health, trained on massive datasets. But does this translate to better prediction for specific outcomes, like AD? Preprint by @twktan :…

2

35

78

38

11.0K

Shaoshi Zhang Retweeted

T

Thomas Yeo@bttyeo · Nov 20

🚨 Predicting Alzheimer's Progression 🚨 A thread 🧵 1/ Accurate prediction of Alzheimer’s progression is critical for early intervention. How can we make predictions more precise and generalizable? 🧠✨ 📝 Read the preprint led by @ChenZhang_NUS : doi.org/10.1101/2024.1…

1

13

48

26

5.0K