Back

Contrastive Transformer-Driven Discovery of Temporal Hemodynamic Subphenotypes in Cardiac Surgery Patients

Desman, J. M.; Sabounchi, M.; Oh, W.; Kumar, G.; Shaikh, A.; Gupta, R.; Gidwani, U.; Manasia, A.; Varghese, R.; Oropello, J.; Smith, G.; Kia, A.; Timsina, P.; Kaplan, B.; Shetreat-Klein, A.; Glicksberg, B.; Legrand, M.; Khanna, A. K.; Kellum, J. A.; Kovatch, P.; Kohli-Seth, R.; Charney, A. W.; Reich, D.; Nadkarni, G. N.; Sakhuja, A.

2026-03-30 intensive care and critical care medicine
10.64898/2026.03.27.26349519 medRxiv
Show abstract

Cardiac surgery patients experience rapidly evolving hemodynamics in early post-operative period requiring intensive support. Identifying hemodynamic subphenotypes from these data can inform personalized management. Using 24-hour high-resolution physiologic and treatment data from 6,630 MIMIC-IV and 1,963 SICdb patients, we trained a transformer encoder with a reconstruction-contrastive objective to derive patient-level embeddings capturing multivariate temporal dynamics within first 24h of ICU stay and compared them against those generated by dynamic time warping (DTW). Spectral clustering uncovered three reproducible hemodynamic subphenotypes. Compared with subphenotype 1, subphenotype 3 received more IV fluids, vasopressors, inotropes, and exhibited higher in-hospital mortality (OR 5.85, 95 % CI 2.43-14.13), longer ICU stay (7.12 days, 95% CI: 5.52-8.73) and hospitalization (8.86 days, 95% CI: 6.57-11.16). DTW derived subphenotypes had weaker prognostic separation. Thus, contrastive-transformer framework identified more clinically meaningful temporal hemodynamic subphenotypes that may optimize post-operative risk stratification and inform personalized management.

Matching journals

The top 7 journals account for 50% of the predicted probability mass.

1
eBioMedicine
130 papers in training set
Top 0.1%
12.3%
2
Nature
575 papers in training set
Top 3%
10.1%
3
Science Translational Medicine
111 papers in training set
Top 0.1%
9.1%
4
Nature Machine Intelligence
61 papers in training set
Top 0.3%
7.2%
5
Advanced Science
249 papers in training set
Top 3%
6.4%
6
npj Digital Medicine
97 papers in training set
Top 0.9%
4.9%
7
Nature Biomedical Engineering
42 papers in training set
Top 0.1%
4.9%
50% of probability mass above
8
Nature Medicine
117 papers in training set
Top 0.8%
3.6%
9
Scientific Reports
3102 papers in training set
Top 41%
3.1%
10
eLife
5422 papers in training set
Top 36%
2.1%
11
Nature Communications
4913 papers in training set
Top 47%
2.1%
12
iScience
1063 papers in training set
Top 12%
1.9%
13
Science Advances
1098 papers in training set
Top 17%
1.7%
14
Nature Methods
336 papers in training set
Top 4%
1.7%
15
Proceedings of the National Academy of Sciences
2130 papers in training set
Top 36%
1.3%
16
Bioinformatics
1061 papers in training set
Top 8%
1.2%
17
Journal of Biomedical Informatics
45 papers in training set
Top 1%
1.1%
18
PLOS Computational Biology
1633 papers in training set
Top 21%
0.9%
19
Cell Systems
167 papers in training set
Top 10%
0.9%
20
Cell Reports Medicine
140 papers in training set
Top 6%
0.9%
21
Patterns
70 papers in training set
Top 2%
0.9%
22
Cell Research
49 papers in training set
Top 2%
0.8%
23
Medical Image Analysis
33 papers in training set
Top 0.9%
0.8%
24
Genome Medicine
154 papers in training set
Top 7%
0.8%
25
PLOS ONE
4510 papers in training set
Top 68%
0.7%
26
Communications Biology
886 papers in training set
Top 24%
0.7%
27
Imaging Neuroscience
242 papers in training set
Top 3%
0.7%
28
Nature Computational Science
50 papers in training set
Top 2%
0.7%
29
Science
429 papers in training set
Top 20%
0.7%
30
Cell Reports
1338 papers in training set
Top 35%
0.6%