# Decompensation Prediction on MIMIC-III (test)

Source: [Decompensation Prediction on MIMIC-III (test)](<https://www.sota2.com/research/sota/decompensation-prediction-on-mimic-iii-test>)

Collection index: [llms.txt](<https://www.sota2.com/research/llms.txt>)

Data updated: 2026-02-26T05:25:38.533Z

Evaluation of model performance in predicting patient decompensation events using the test split of the MIMIC-III clinical database.

A research result applies to its stated dataset, split, metric and evaluation configuration. Compare like-for-like settings, preserve the metric's unit and direction, and distinguish paper-reported results from independent reproduction. The best result recorded here is not a claim of exhaustive, current global coverage.

| Scope | Recorded value |
| --- | --- |
| Task | Decompensation Prediction |
| Dataset (including any stated split) | MIMIC-III (test) |
| Metric | AUPRC |
| Unit | PERCENT |
| Direction | Higher is better |
| Results loaded | 8 of 8 |
| Coverage | All results currently returned for this benchmark |

Evaluation settings may differ between results. The original papers and recorded configurations are needed to establish comparability. A shared metric name alone does not establish an identical protocol.

## Metrics

- [AUPRC](<https://www.sota2.com/research/sota/decompensation-prediction-on-mimic-iii-test.md?metric=AUPRC>): Higher is better; unit: PERCENT.

- [Timestep Recall](<https://www.sota2.com/research/sota/decompensation-prediction-on-mimic-iii-test.md?metric=Timestep%20Recall>): Higher is better; unit: PERCENT.

- [Event Recall](<https://www.sota2.com/research/sota/decompensation-prediction-on-mimic-iii-test.md?metric=Event%20Recall>): Higher is better; unit: PERCENT.

## Recorded results

| Method | AUPRC | Result date | Report paper |
| --- | --- | --- | --- |
| Temporal label smoothing | 35.5 | Not recorded | 68fa6e781e209131cea4019f |
| Multi-horizon | 34.9 | Not recorded | 68fa6e781e209131cea4019f |
| Cross-entropy | 34.5 | Not recorded | 68fa6e781e209131cea4019f |
| Weighted CE | 34.5 | Not recorded | 68fa6e781e209131cea4019f |
| Focal loss | 34.5 | Not recorded | 68fa6e781e209131cea4019f |
| Landmarking | 34 | Not recorded | 68fa6e781e209131cea4019f |
| Label smoothing | 33.9 | Not recorded | 68fa6e781e209131cea4019f |
| TCSR | 28.6 | Not recorded | 68fa6e781e209131cea4019f |

### Temporal label smoothing

AUPRC: 35.5. Result ID: 6990237e3a6c314849fe6032.

[Reporting paper](<https://www.sota2.com/research/paper/68fa6e781e209131cea4019f.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa6e781e209131cea4019f.md>)

[Implementation](<https://anonymous.4open.science/r/tls/>)

Recorded evaluation configuration: {"Training objective":"Temporal label smoothing"}. This record does not assert independent reproduction.

### Multi-horizon

AUPRC: 34.9. Result ID: 6990237e3a6c314849fe602f.

[Reporting paper](<https://www.sota2.com/research/paper/68fa6e781e209131cea4019f.md>)

Recorded evaluation configuration: {"Training objective":"Multi-horizon"}. This record does not assert independent reproduction.

### Cross-entropy

AUPRC: 34.5. Result ID: 6990237e3a6c314849fe602b.

[Reporting paper](<https://www.sota2.com/research/paper/68fa6e781e209131cea4019f.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa65e11e209131cea0d18f.md>)

Recorded evaluation configuration: {"Training objective":"Cross-entropy"}. This record does not assert independent reproduction.

### Weighted CE

AUPRC: 34.5. Result ID: 6990237e3a6c314849fe602c.

[Reporting paper](<https://www.sota2.com/research/paper/68fa6e781e209131cea4019f.md>)

Recorded evaluation configuration: {"Training objective":"Weighted CE"}. This record does not assert independent reproduction.

### Focal loss

AUPRC: 34.5. Result ID: 6990237e3a6c314849fe602d.

[Reporting paper](<https://www.sota2.com/research/paper/68fa6e781e209131cea4019f.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa640c1e209131cea02556.md>)

[Implementation](<https://github.com/facebookresearch/Detectron>)

Recorded evaluation configuration: {"Training objective":"Focal loss"}. This record does not assert independent reproduction.

### Landmarking

AUPRC: 34. Result ID: 6990237e3a6c314849fe6030.

[Reporting paper](<https://www.sota2.com/research/paper/68fa6e781e209131cea4019f.md>)

Recorded evaluation configuration: {"Training objective":"Landmarking"}. This record does not assert independent reproduction.

### Label smoothing

AUPRC: 33.9. Result ID: 6990237e3a6c314849fe602e.

[Reporting paper](<https://www.sota2.com/research/paper/68fa6e781e209131cea4019f.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa62331e209131ce9f8333.md>)

Recorded evaluation configuration: {"Training objective":"Label smoothing"}. This record does not assert independent reproduction.

### TCSR

AUPRC: 28.6. Result ID: 6990237e3a6c314849fe6031.

[Reporting paper](<https://www.sota2.com/research/paper/68fa6e781e209131cea4019f.md>)

Recorded evaluation configuration: {"Training objective":"TCSR"}. This record does not assert independent reproduction.

Page 1 of 1. This is an index page, not the full collection.