Back

Transformer-based artificial intelligence on single-cell clinical data for homeostatic mechanism inference and rational biomarker discovery

2025-03-25 hematology Title + abstract only
View on medRxiv
Show abstract

Artificial intelligence (AI) applied to single-cell data has the potential to transform our understanding of biological systems by revealing patterns and mechanisms that simpler traditional methods miss. Here, we develop a general-purpose, interpretable AI pipeline consisting of two deep learning models: the Multi- Input Set Transformer++ (MIST) model for prediction and the single-cell FastShap model for interpretability. We apply this pipeline to a large set of routine clinical data containing ...

Predicted journal destinations

1
Scientific Reports
701 training papers
Top 0.6% 25.2%
2
Nature Communications
483 training papers
Top 11% 8.7%
3
PLOS Computational Biology
141 training papers
Top 1% 8.7%
4
PLOS ONE
1737 training papers
Top 70% 6.7%
5
Blood Advances
16 training papers
Top 0.5% 6.3%
6
Nature Genetics
72 training papers
Top 3% 5.6%
7
Science Advances
52 training papers
#1 4.3%
8
npj Digital Medicine
85 training papers
Top 8% 3.1%
9
eLife
262 training papers
Top 18% 2.2%
10
The American Journal of Human Genetics
77 training papers
Top 6% 1.7%
11
Proceedings of the National Academy of Sciences
100 training papers
Top 7% 1.7%
12
Communications Medicine
63 training papers
Top 1% 1.7%
13
Bioinformatics
24 training papers
Top 1% 1.3%
14
Cell Genomics
34 training papers
Top 6% 1.1%
15
PLOS Digital Health
88 training papers
Top 15% 1.1%
16
Science Translational Medicine
40 training papers
Top 3% 1.1%
17
JCI Insight
63 training papers
Top 6% 1.1%
18
iScience
74 training papers
Top 9% 1.1%
19
Frontiers in Immunology
140 training papers
Top 10% 1.1%
20
PLOS Genetics
39 training papers
Top 7% 1.1%
21
Statistics in Medicine
17 training papers
Top 0.4% 0.8%
22
Journal of Clinical Investigation
50 training papers
Top 7% 0.5%
23
Nature
58 training papers
Top 13% 0.5%
24
Cell Reports Medicine
49 training papers
Top 9% 0.5%
25
Communications Biology
36 training papers
Top 8% 0.5%
26
Cell
28 training papers
Top 3% 0.5%
27
Frontiers in Medicine
99 training papers
Top 30% 0.5%
28
Nature Medicine
88 training papers
Top 24% 0.5%