Machine Learning Models Uncover Subphenotypes of AKI With Unique Signatures That Associate With Differing Clinical Outcomes
Bibliographic record
Abstract
Background: Acute kidney injury (AKI) is defined through serum creatinine and urine output metrics. However, these markers do not capture the complexity of AKI and do not fully inform on the future risk of kidney and clinical events. Methods: We evaluated clinical and biomarker data from AKI patients during the acute hospitalization from ASSESS-AKI via three machine learning algorithms to uncover different AKI composites. We compared key characteristics within each subphenotype via classic statistics and then examined the time to event for kidney events (CKD incidence and progression), cardiovascular events, and death by subphenotype. Results: We included 748 AKI patients. The mean age (± SD) was 64 (13) years, 67.9% were men, and the median follow-up was 4.8 years. Patients with AKI subphenotype 1 (‘cardiorenal injury’, N=181) were characterized by prevalent CVD (78%, P<0.001) and the highest levels of KIM-1, urinary IL-18, and Troponin T. Subphenotype 2 (‘benign’, N=250) was comprised of individuals with a low prevalence of comorbid conditions and high uromodulin levels, a marker of tubular repair. AKI subphenotype 3 (‘cardiorenal inflammation, N=159) comprised patients with markedly high levels of pro-BNP, TNFRs and low kidney injury (KIM-1, NGAL). Finally, patients subphenotype 4 (‘sepsis-AKI’, N=158) had high rates of infections and dialysis-requiring AKI. These patients had the highest levels of vascular/kidney (YKL-40, MCP-1), and injury activity. AKI subphenotype 3 and 4 were independently associated with a higher risk of death: adjusted hazard ratios (aHR) of 2.9 (95% CI: 1.8 - 4.6, p<0.001) and 1.6 (1.01 - 2.6, p=0.04), respectively. Subphenotype 3 was also independently associated with triple the risk of CKD outcomes (aHR: 2.6, CI: 1.6 - 4.2) and CVD events (aHR: 2.6, CI: 1.6 - 4.1). Conclusions: We discovered four novel and clinically meaningful AKI subphenotypes that inform on potential pathway abnormalities that associate with differing risks for long-term events. We found a new role for biomarkers when they are evaluated in an agnostic fashion, which can serve to advance precision medicine in AKI care. Funding: NIDDK Support
Fetched live from OpenAlex and de-inverted. Abstracts are not stored in this database: the inverted indexes are 8.6 GB of the frame’s 9.3 GB of text, and the host has 13 GB free.
How this classification was reachedexpand
Full frame distilled prediction
Teacher imitationNot calibrated prevalence, not ground truth. Human validation pending. Learned from the 10,348 direct Codex labels and 10,348 direct Gemma labels. Candidate is the union of thresholded teacher heads; consensus is their intersection. These outputs are machine_predicted_unvalidated and are not human labels or direct frontier model labels.
Codex and Gemma teacher scores by category
| Category | Codex | Gemma |
|---|---|---|
| Metaresearch | 0.001 | 0.000 |
| Meta-epidemiology (narrow) | 0.000 | 0.000 |
| Meta-epidemiology (broad) | 0.002 | 0.001 |
| Bibliometrics | 0.000 | 0.000 |
| Science and technology studies | 0.000 | 0.001 |
| Scholarly communication | 0.000 | 0.000 |
| Open science | 0.000 | 0.000 |
| Research integrity | 0.000 | 0.001 |
| Insufficient payload (model declined to judge) | 0.000 | 0.000 |
Machine scores (provisional)
The two teacher heads of the student model, read on this work. A score orders the frame for review; it never asserts a category, and the validation status ships verbatim with every row.
Baseline scores from an immature model (maturity gate not passed, 7 training rounds). Scores rank; they never assert a category.
score_only:v0-immature-baseline · verbatim from the scoring run: score_only means the number may rank works, and no category label ships from itClassification
machine, unvalidatedMachine predicted; a candidate call from one teacher head, not a consensus.
How this classification was reached, model by model and score by score, is at the end of the page under "How this classification was reached".