Sign in

Lijun An

@anlijuncn.bsky.social
202 followers 80 following 28 posts

Postdoc@Jacob Vogel ⬅️ PhD@Thomas Yeo. Neurodegenerative Disease, Brain Imaging, Machine Learning, Multi-omics anlijun.cn

PostsRepliesMedia
Lijun An @anlijuncn.bsky.social · 10/09/2026
📄 Preprint: medrxiv.org/content/10.6... Feedback welcome! #Proteomics #Biomarkers #Neurology /end
medrxiv.org
000 0
Lijun An @anlijuncn.bsky.social · 10/09/2026
The fix: split first. Keep feature selection, preprocessing & tuning inside the training data, and validate externally. Two open tools to catch leakage: 🔧 Leakly checks your pipeline's code: pypi.org/project/Leak... 🔧 Plumber screens manuscripts: demonlab-biofinder.github.io/Plumber/
pypi.org
Leakly
Leakage checks for machine-learning pipelines using permutation tests.
110 0
Lijun An @anlijuncn.bsky.social · 10/09/2026
We quantified this in 500 simulations + GNPC (N=7,548; AD/PD/FTD; 5 cohorts), 6 classifiers: 📈 Within-cohort AUC: +11 points on average (often +25) ⚠️ No true signal: some leaky runs still approached AUC 0.8 📉 External cohorts: leaky models ~10 points worse
100 0
Lijun An @anlijuncn.bsky.social · 10/09/2026
In a typical pipeline, proteins are selected on the full cohort because they differ between cases & controls. An ML model is then cross-validated on how well those same proteins separate cases from controls, in the same data. It's circular, so performance is inflated.
100 0
Lijun An @anlijuncn.bsky.social · 10/09/2026
🧵 Struggling to replicate a proteomics biomarker study? Our new preprint quantifies one common cause: data leakage. With NO true signal, some runs with data leakage still reached AUCs near 0.8. w/ @jwvogel.bsky.social, Caitlin Finney, Artur Shvetcov & @neuroproteome.bsky.social
110 0
Reposted by Lijun An
biofinder.bsky.social @biofinder.bsky.social · 15/05/2026
New paper in Nature Aging led by PhD student Lina Lu! We mapped how APOE ε4 and APOE ε2 shape molecular changes in blood and cerebrospinal fluid across Alzheimer’s disease. www.nature.com/articles/s43...
nature.com
Proteomic signatures of the APOE ε4 and APOE ε2 genetic variants and Alzheimer’s disease - Nature Aging
Apolipoprotein E (APOE) is the strongest genetic influence in Alzheimer’s disease (AD). Compared to the most frequent allele, ε3, the ε4 allele increases AD risk, and the ε2 allele is protective. Here...
122 0
Lijun An @anlijuncn.bsky.social · 15/05/2026
If you want to learn proteomics signatures of APOE genetic variants on a massive sample from multi-chort, you should not miss this tour de force! Huge congrats to Lu Lina and Niklas!
010 0
Reposted by Lijun An
Evan Gordon @gordonneuro.bsky.social · 22/04/2026
I always assumed that brain function had to line up with cytoarchitectonics. It turns out I was wrong. Human cortex, especially PFC, is tiled by chains of functional patches that subdivide and interlink architectonic areas into parallel processing streams. www.biorxiv.org/content/10.6...
14234122 0
Reposted by Lijun An
Healthcare in Europe @healthcare-europe.bsky.social · 10/04/2026
Researchers led by Jacob Vogel and @anlijuncn.bsky.social from 🇸🇪 @lund-university.bsky.social, with @biofinder.bsky.social and @neuroproteome.bsky.social have developed an #AI model to detect different neurodegenerative diseases like #Alzheimers and Lewy body disease from a single blood sample.
healthcare-in-europe.com
New AI model detects multiple brain diseases from a single blood sample
The symptom profiles of different neurodegenerative diseases often overlap, and diagnosing age-related cognitive symptoms is complex. A patient may have multiple overlapping disease processes in the brain at the same time.
011 0
Reposted by Lijun An
Thomas Yeo @bttyeo.bsky.social · 09/04/2026
We develop a new TMS targeting algorithm and test it in an open label trial in a treatment-resistant depression population with high comorbidities. Preprints by @rubykong92.bsky.social Phern-Chern Tor 1. doi.org/10.1101/2025... 2. doi.org/10.64898/202... Our new approach ...
15423 0
Reposted by Lijun An
Jake Vogel @jwvogel.bsky.social · 06/04/2026
Incredibly excited that the lab's first papers is now published! We use AI to simultaneously predict multiple neurodegenerative disease diagnoses from plasma proteomics. Congrats @anlijuncn.bsky.social ! #MedSky #neuroskyence #neurosky #alzsky #compneuro #ai #datascience #bioinformatics #neurology
0133 0
Reposted by Lijun An
Elvisha Dhamala @elvisha.bsky.social · 25/03/2026
Our latest work looking at the neuroanatomical basis of impulsivity in youth is out now in Molecular Psychiatry!
nature.com
Neuroanatomy reflects individual variability in impulsivity in youth - Molecular Psychiatry
Molecular Psychiatry - Neuroanatomy reflects individual variability in impulsivity in youth
24025 0
Lijun An @anlijuncn.bsky.social · 06/04/2026
Finally, great thanks to all of our wonderful collaborators: Romina Zendehdel, Zijian Dong, Bart Smets, Rowan Saloner, Shinya Tasaki, Ying Xu, Varsha Krish, Farhad Imam, Danielle van Westen, and Christopher D. Whelan.
000 0
Lijun An @anlijuncn.bsky.social · 06/04/2026
Massive thanks to the fantastic @biofinder.bsky.social team—Shorena Janelidze, Erik Stomrud, Sebastian Palmqvist, Rik Ossenkoppele, Niklas Mattsson-Carlgren, and Oskar Hansson for sharing these exceptional data and supporting this project every step of the way.
100 0
Lijun An @anlijuncn.bsky.social · 06/04/2026
I would especially like to thank my mentor, @jwvogel.bsky.social, for guiding and supporting this work from day one. I am also deeply grateful to our amazing team: especially @alexapb.bsky.social, Ines Hristovska, Gabriele Vilkaite, and @xiaoyucaly.bsky.social.
110 0
Lijun An @anlijuncn.bsky.social · 06/04/2026
This work would not have been possible without the world’s largest neuroproteomics consortium, GNPC (@neuroproteome.bsky.social), the exceptional BioFINDER-2 cohort (@biofinder.bsky.social), with its rich fluid and imaging biomarker, and the outstanding computational support from alzheimersdata.org
alzheimersdata.org
AD Data Initiative | Home
Originally designed to award a single $1M prize, the competition expanded to name two winners—awarding $1M to each team and doubling the total prize amount to $2M.
100 0
Lijun An @anlijuncn.bsky.social · 06/04/2026
Although the model was trained on baseline data only, its baseline predictions stratified longitudinal cognitive decline for up to 20 years in GNPC, and successfully triaged MCI patients in the BioFINDER-2 memory clinic cohort.
100 0
Lijun An @anlijuncn.bsky.social · 06/04/2026
Beyond diagnostic accuracy, we now show more clinical value of our model, ProtAIDe-Dx. Compared to best possible clinical biomarker panels, ProtAIDe-Dx still adds significant value to PD and FTD diagnosis.
100 0
Lijun An @anlijuncn.bsky.social · 06/04/2026
Our work is now out in @natmed.nature.com www.nature.com/articles/s41... We developed a state-of-the-art AI-proteomics model for diagnosing multiple neurodegenerative diseases from blood plasma. ProtAIDe-Dx now outperformed both classic machine learning models and even foundation models like TabPFN!
143 0
Reposted by Lijun An
Mark A. Hanson @hansonmark.bsky.social · 11/11/2025
We wrote the Strain on scientific publishing to highlight the problems of time & trust. With a fantastic group of co-authors, we present The Drain of Scientific Publishing: a 🧵 1/n Drain: arxiv.org/abs/2511.04820 Strain: direct.mit.edu/qss/article/... Oligopoly: direct.mit.edu/qss/article/...
A table showing profit margins of major publishers. A snippet of text related to this table is below.

1. The four-fold drain
1.1 Money
Currently, academic publishing is dominated by profit-oriented, multinational companies for
whom scientific knowledge is a commodity to be sold back to the academic community who
created it. The dominant four are Elsevier, Springer Nature, Wiley and Taylor & Francis,
which collectively generated over US$7.1 billion in revenue from journal publishing in 2024
alone, and over US$12 billion in profits between 2019 and 2024 (Table 1A). Their profit
margins have always been over 30% in the last five years, and for the largest publisher
(Elsevier) always over 37%.
Against many comparators, across many sectors, scientific publishing is one of the most
consistently profitable industries (Table S1). These financial arrangements make a substantial
difference to science budgets. In 2024, 46% of Elsevier revenues and 53% of Taylor &
Francis revenues were generated in North America, meaning that North American
researchers were charged over US$2.27 billion by just two for-profit publishers. The
Canadian research councils and the US National Science Foundation were allocated US$9.3
billion in that year.A figure detailing the drain on researcher time.

1. The four-fold drain

1.2 Time
The number of papers published each year is growing faster than the scientific workforce,
with the number of papers per researcher almost doubling between 1996 and 2022 (Figure
1A). This reflects the fact that publishers’ commercial desire to publish (sell) more material
has aligned well with the competitive prestige culture in which publications help secure jobs,
grants, promotions, and awards. To the extent that this growth is driven by a pressure for
profit, rather than scholarly imperatives, it distorts the way researchers spend their time.
The publishing system depends on unpaid reviewer labour, estimated to be over 130 million
unpaid hours annually in 2020 alone (9). Researchers have complained about the demands of
peer-review for decades, but the scale of the problem is now worse, with editors reporting
widespread difficulties recruiting reviewers. The growth in publications involves not only the
authors’ time, but that of academic editors and reviewers who are dealing with so many
review demands.
Even more seriously, the imperative to produce ever more articles reshapes the nature of
scientific inquiry. Evidence across multiple fields shows that more papers result in
‘ossification’, not new ideas (10). It may seem paradoxical that more papers can slow
progress until one considers how it affects researchers’ time. While rewards remain tied to
volume, prestige, and impact of publications, researchers will be nudged away from riskier,
local, interdisciplinary, and long-term work. The result is a treadmill of constant activity with
limited progress whereas core scholarly practices – such as reading, reflecting and engaging
with others’ contributions – is de-prioritized. What looks like productivity often masks
intellectual exhaustion built on a demoralizing, narrowing scientific vision.A table of profit margins across industries. The section of text related to this table is below:

1. The four-fold drain
1.1 Money
Currently, academic publishing is dominated by profit-oriented, multinational companies for
whom scientific knowledge is a commodity to be sold back to the academic community who
created it. The dominant four are Elsevier, Springer Nature, Wiley and Taylor & Francis,
which collectively generated over US$7.1 billion in revenue from journal publishing in 2024
alone, and over US$12 billion in profits between 2019 and 2024 (Table 1A). Their profit
margins have always been over 30% in the last five years, and for the largest publisher
(Elsevier) always over 37%.
Against many comparators, across many sectors, scientific publishing is one of the most
consistently profitable industries (Table S1). These financial arrangements make a substantial
difference to science budgets. In 2024, 46% of Elsevier revenues and 53% of Taylor &
Francis revenues were generated in North America, meaning that North American
researchers were charged over US$2.27 billion by just two for-profit publishers. The
Canadian research councils and the US National Science Foundation were allocated US$9.3
billion in that year.The costs of inaction are plain: wasted public funds, lost researcher time, compromised
scientific integrity and eroded public trust. Today, the system rewards commercial publishers
first, and science second. Without bold action from the funders we risk continuing to pour
resources into a system that prioritizes profit over the advancement of scientific knowledge.
8644453 0
Reposted by Lijun An
Thomas Yeo @bttyeo.bsky.social · 20/11/2024
🚨 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 @chen-zhang.bsky.social : doi.org/10.1101/2024...
12311 0
Reposted by Lijun An
Thomas Yeo @bttyeo.bsky.social · 04/11/2025
Preprint is now published! doi.org/10.1002/hbm.... Thanks to co-authors @chen-zhang.bsky.social @anlijuncn.bsky.social @csabaorban.bsky.social
doi.org
Cross‐Dataset Evaluation of Dementia Longitudinal Progression Prediction Models
We introduced a simplified variant of FROG, the winning algorithm of the TADPOLE challenge, for predicting dementia progression using multimodal longitudinal data. Combining the longitudinal-to-cross...
1143 0
Lijun An @anlijuncn.bsky.social · 22/07/2025
Check out lab's latest preprint for MRI super resolution!!!
030 0
Reposted by Lijun An
Jake Vogel @jwvogel.bsky.social · 17/07/2025
Incredibly excited for this new work from our lab. We test the potential of AI-based neurodegenerative disease diagnostics using plasma proteomics data from n>17,000 people, led by the brilliant and indefatigable @anlijuncn.bsky.social Check it out!👇
032 0
Reposted by Lijun An
Thomas Yeo @bttyeo.bsky.social · 17/07/2025
1/11 Excited to share our @Naturestudy led by @leonooi.bsky.social @csabaorban.bsky.social @shaoshiz.bsky.social AI performance is known to scale with logarithm of sample size (Kaplan 2020), but in many domains, sample size can be # participants or # measurements... doi.org/10.1038/s415...
317182 0
Lijun An @anlijuncn.bsky.social · 15/07/2025
dsdsd
000 0
Lijun An @anlijuncn.bsky.social · 15/07/2025
[10/10] Finally, thanks to all our best collaborators: Bart Smets, Rowan Saloner, Shinya Tasaki, Ying Xu, Varsha Krish, Farhad Imam, Danielle van Westen, Christopher D. Whelan!
100 0
Lijun An @anlijuncn.bsky.social · 15/07/2025
[9/10] And massive thanks to the coolest ‪@biofinder.bsky.social‬ team—Shorena Janelidze, Erik Stomrud, Sebastian Palmqvist, ‪@rikossenkoppele.bsky.social‬, Niklas Mattsson-Carlgren, and Oskar Hansson —for sharing the awesome data and supporting this project every step of the way.
100 0
Lijun An @anlijuncn.bsky.social · 15/07/2025
[8/10] Endless thanks to ‪@jwvogel.bsky.social‬ for guiding and supporting this work from day one. To our amazing team, especially Alexa Pichet Binette, Ines Hristovska, Gabriele Vilkaite, and @xiaoyucaly.bsky.social‬, for their enormous support.
110 0
Lijun An @anlijuncn.bsky.social · 15/07/2025
[7/10] ProtAIDe-Dx is a step toward the development of scalable, minimally invasive, and multi-disease diagnostic tools for neurodegenerative diseases. We hope our work can establish a benchmark for AI-driven proteomics tools, paving the way for precision medicine in these diseases.
100 0
Lijun An @anlijuncn.bsky.social · 15/07/2025
[6/10] We built a proof-of-concept diagnostic report with ProtAIDe-Dx, visualizing diagnostic probabilities across conditions and highlighting proteins that contributed to the individual’s prediction and traits linked to those proteins, enabling a biologically interpretable diagnosis.[Figure 5]
100 0
Lijun An @anlijuncn.bsky.social · 15/07/2025
[5/10] Could ProtAIDe-Dx aid real-world clinical application? In a memory clinic cohort, ProtAIDe-Dx significantly improved differential diagnosis (especially PD and CVD) when combined with accessible clinical biomarkers (MMSE, MRI, plasma p-tau217 & NfL). (Figure 4D).
100 0
Lijun An @anlijuncn.bsky.social · 15/07/2025
[4/10] ProtAIDe-Dx identified a novel and compact set of top predictive proteins. Notably, OMG and Histone H1-2 (H1-2) distinguished controls from all neurodegenerative conditions, and ubiquitin (UBB) was a discriminative protein for FTD. [Figure 3A].
100 0
Lijun An @anlijuncn.bsky.social · 15/07/2025
[3/10] The models’ diagnostic probabilities highlighted subgroups of patients with proteomic patterns more similar to a different condition, e.g., AD patients resembling stroke patients, indicating possible misdiagnosis, potential co-pathology, or may even indicate distinct etiologies. [Figure 2A].
100 0
Lijun An @anlijuncn.bsky.social · 15/07/2025
[2/10] We developed a multi-task model, ProtAIDe-Dx, to provide simultaneous probabilistic diagnosis across six conditions associated with dementia in aging (Control, AD, PD, FTD, ALS, and Stroke/TIA). CVed balanced ranged from 69%-96% and AUCs > 79% across all conditions. [Figure 1B]
100 0
Lijun An @anlijuncn.bsky.social · 15/07/2025
[1/10] For the first time, we can answer this question with the world's largest neurodegeneration proteomics consortium, GNPC @neuroproteome.bsky.social #GNPC #Neurodegeneration #Proteomics #OpenScience #ADRD #Biomarkers #Parkinsons #ALS #FTD #NatureMedicine #NatureAging #SomaScan #Plasma #CSF
100 0
Lijun An @anlijuncn.bsky.social · 15/07/2025
Can AI reveal risk & co-pathology of multiple neurodegenerative diseases from a single blood sample? We explored AI-based diagnostic power on high rank plasma proteomics (N=17,170). www.medrxiv.org/content/10.1... #neuroskyence #neurosky #Alzheimer #compneuro #AI #datascience #neurology
medrxiv.org
Benchmarking the AI-based diagnostic potential of plasma proteomics for neurodegenerative disease in 17,170 people
Co-pathology is a common feature of neurodegenerative diseases that complicates diagnosis, treatment and clinical management. However, sensitive, specific and scalable biomarkers for in vivo pathologi...
1115 0
Lijun An @anlijuncn.bsky.social · 02/05/2025
Check out Xiaoyu’s fantastic work!!
021 0
Reposted by Lijun An
Yu Xiao (肖燏) @xiaoyucaly.bsky.social · 24/04/2025
🧵15/ Huge thanks to our amazing team and coauthors! Endless thanks to @jwvogel.bsky.social for guiding and supporting this work from day one. To our amazing team DeMON lab, especially @anlijuncn.bsky.social for enormous support.
121 0
Reposted by Lijun An
Shaoshi Zhang @shaoshiz.bsky.social · 11/04/2025
Check our latest preprint led by the amazing @tianchu.bsky.social and @tianfang.bsky.social where we speed up the tedious parameter optimization process for biophysical modelling
011 0
Reposted by Lijun An
Lianglong Sun @longonga.bsky.social · 04/04/2025
Happy to share that our article “Human lifespan changes in the brain’s functional connectome” is now published online at Nature Neuroscience @natureneuro.bsky.social ! Many thanks to all collaborators & data contributors, and the editor team & reviewers! www.nature.com/articles/s41...
nature.com
Human lifespan changes in the brain’s functional connectome - Nature Neuroscience
Sun et al. report human lifespan changes in the brain’s functional connectome in 33,250 individuals, which highlights critical growth milestones and distinct maturation patterns and offers a normative...
23519 0
Reposted by Lijun An
Thomas Yeo @bttyeo.bsky.social · 11/04/2025
While the world burns, we cook up a new preprint! doi.org/10.1101/2025... Biophysical modeling is a key tool to derive mechanistic insights into the brain. These models are governed by biologically meaningful parameters (unlike artificial neural networks), but the dirty secret ... 1/N
110839 0
Reposted by Lijun An
Thomas Yeo @bttyeo.bsky.social · 26/03/2025
Updated preprint for those who might be interested: doi.org/10.1101/2024...
doi.org
Longer scans boost prediction and cut costs in brain-wide association studies
A pervasive dilemma in brain-wide association studies (BWAS) is whether to prioritize functional MRI (fMRI) scan time or sample size. We derive a theoretical model showing that individual-level phenot...
03417 0
Lijun An @anlijuncn.bsky.social · 26/03/2025
Excellent work by Ruby!!
020 0
Reposted by Lijun An
Hu Chuan-Peng is on paternity leave @hcp4715.bsky.social · 11/12/2024
a cool new study "established in India & Tanzania, with appropriate training, structured teams, & daily automated analysis & feedback, non-specialists can reliably collect EEG data alongside various survey & assessments w/ consistently high throughput & quality. " www.biorxiv.org/content/10.1...
biorxiv.org
EEG data quality in large scale field studies in India and Tanzania
There is a growing imperative to understand the neurophysiological impact of our rapidly changing and diverse technological, social, chemical, and physical environments. To untangle the multidimension...
021 0
Reposted by Lijun An
Emily Finn @esfinn.bsky.social · 11/12/2024
This paper sets up a bit of a straw man in that I don't think most people who use movies and stories as fMRI stimuli assume that all movies will (or should) evoke the same response. In a naturalistic neuroimaging expt, the movie *is* the task...
biorxiv.org
Between-movie variability severely limits generalizability of “naturalistic” neuroimaging
“Naturalistic imaging” paradigms, where participants watch movies during fMRI, have gained popularity over the past two decades. Many movie-watching studies measure inter-subject correlation (ISC), wh...
10717 0
Reposted by Lijun An
Thomas Yeo @bttyeo.bsky.social · 20/11/2024
🚨 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.bsky.social : doi.org/10.1101/2024...
24620 0