# Multi-label Toxin Prediction on Hospital Poly-intoxication Dataset

Source: [Multi-label Toxin Prediction on Hospital Poly-intoxication Dataset](<https://www.sota2.com/research/sota/multi-label-toxin-prediction-on-hospital-poly-intoxication-dataset>)

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

Data updated: 2026-04-01T08:26:14.382Z

Evaluation of multi-label classification models on the Hospital Poly-intoxication Dataset to predict the presence of multiple toxins simultaneously. Performance is typically measured using micro/macro averaged F1, Precision, and Recall.

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 | Multi-label toxin prediction |
| Dataset (including any stated split) | Hospital Poly-intoxication Dataset |
| Metric | F1 Micro |
| Unit | PERCENT |
| Direction | Higher is better |
| Results loaded | 11 of 11 |
| 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

- [F1 Micro](<https://www.sota2.com/research/sota/multi-label-toxin-prediction-on-hospital-poly-intoxication-dataset.md?metric=F1%20Micro>): Higher is better; unit: PERCENT.

- [F1 Macro](<https://www.sota2.com/research/sota/multi-label-toxin-prediction-on-hospital-poly-intoxication-dataset.md?metric=F1%20Macro>): Higher is better; unit: PERCENT.

- [Recall Micro](<https://www.sota2.com/research/sota/multi-label-toxin-prediction-on-hospital-poly-intoxication-dataset.md?metric=Recall%20Micro>): Higher is better; unit: PERCENT.

- [Recall Macro](<https://www.sota2.com/research/sota/multi-label-toxin-prediction-on-hospital-poly-intoxication-dataset.md?metric=Recall%20Macro>): Higher is better; unit: PERCENT.

- [Precision Micro](<https://www.sota2.com/research/sota/multi-label-toxin-prediction-on-hospital-poly-intoxication-dataset.md?metric=Precision%20Micro>): Higher is better; unit: PERCENT.

- [Precision Macro](<https://www.sota2.com/research/sota/multi-label-toxin-prediction-on-hospital-poly-intoxication-dataset.md?metric=Precision%20Macro>): Higher is better; unit: PERCENT.

## Recorded results

| Method | F1 Micro | Result date | Report paper |
| --- | --- | --- | --- |
| DeToxR | 63.71 | Not recorded | 69cca6cc0636e5bf4afcbe5c |
| Qwen3 4B Instruct + SFT | 58.89 | Not recorded | 69cca6cc0636e5bf4afcbe5c |
| XGBoost | 57.12 | Not recorded | 69cca6cc0636e5bf4afcbe5c |
| MLP | 54.39 | Not recorded | 69cca6cc0636e5bf4afcbe5c |
| Qwen3 4B Instruct | 52.63 | Not recorded | 69cca6cc0636e5bf4afcbe5c |
| MedGemma 4B IT | 50.61 | Not recorded | 69cca6cc0636e5bf4afcbe5c |
| Qwen3 30B Instruct (MoE) | 50.14 | Not recorded | 69cca6cc0636e5bf4afcbe5c |
| Llama 3.1 8B Instruct | 49.7 | Not recorded | 69cca6cc0636e5bf4afcbe5c |
| History baseline | 48.81 | Not recorded | 69cca6cc0636e5bf4afcbe5c |
| GPT OSS 20B | 48.76 | Not recorded | 69cca6cc0636e5bf4afcbe5c |
| Qwen3 4B Thinking | 48.22 | Not recorded | 69cca6cc0636e5bf4afcbe5c |

### DeToxR

F1 Micro: 63.71. Result ID: 69ccd69747f34ce924b67d75.

[Reporting paper](<https://www.sota2.com/research/paper/69cca6cc0636e5bf4afcbe5c.md>)

[Method origin paper](<https://www.sota2.com/research/paper/69cca6cc0636e5bf4afcbe5c.md>)

Recorded evaluation configuration: {"Category":"Finetuned models"}. This record does not assert independent reproduction.

### Qwen3 4B Instruct + SFT

F1 Micro: 58.89. Result ID: 69ccd69747f34ce924b67d74.

[Reporting paper](<https://www.sota2.com/research/paper/69cca6cc0636e5bf4afcbe5c.md>)

Recorded evaluation configuration: {"Category":"Finetuned models"}. This record does not assert independent reproduction.

### XGBoost

F1 Micro: 57.12. Result ID: 69ccd69747f34ce924b67d6d.

[Reporting paper](<https://www.sota2.com/research/paper/69cca6cc0636e5bf4afcbe5c.md>)

Recorded evaluation configuration: {"Category":"Baselines"}. This record does not assert independent reproduction.

### MLP

F1 Micro: 54.39. Result ID: 69ccd69747f34ce924b67d6c.

[Reporting paper](<https://www.sota2.com/research/paper/69cca6cc0636e5bf4afcbe5c.md>)

Recorded evaluation configuration: {"Category":"Baselines"}. This record does not assert independent reproduction.

### Qwen3 4B Instruct

F1 Micro: 52.63. Result ID: 69ccd69747f34ce924b67d6e.

[Reporting paper](<https://www.sota2.com/research/paper/69cca6cc0636e5bf4afcbe5c.md>)

Recorded evaluation configuration: {"Category":"Base LLM inference","Zero-shot":"true"}. This record does not assert independent reproduction.

### MedGemma 4B IT

F1 Micro: 50.61. Result ID: 69ccd69747f34ce924b67d73.

[Reporting paper](<https://www.sota2.com/research/paper/69cca6cc0636e5bf4afcbe5c.md>)

Recorded evaluation configuration: {"Category":"Base LLM inference","Zero-shot":"true"}. This record does not assert independent reproduction.

### Qwen3 30B Instruct (MoE)

F1 Micro: 50.14. Result ID: 69ccd69747f34ce924b67d71.

[Reporting paper](<https://www.sota2.com/research/paper/69cca6cc0636e5bf4afcbe5c.md>)

Recorded evaluation configuration: {"Category":"Base LLM inference","Zero-shot":"true"}. This record does not assert independent reproduction.

### Llama 3.1 8B Instruct

F1 Micro: 49.7. Result ID: 69ccd69747f34ce924b67d70.

[Reporting paper](<https://www.sota2.com/research/paper/69cca6cc0636e5bf4afcbe5c.md>)

Recorded evaluation configuration: {"Category":"Base LLM inference","Zero-shot":"true"}. This record does not assert independent reproduction.

### History baseline

F1 Micro: 48.81. Result ID: 69ccd69747f34ce924b67d6b.

[Reporting paper](<https://www.sota2.com/research/paper/69cca6cc0636e5bf4afcbe5c.md>)

Recorded evaluation configuration: {"Category":"Baselines"}. This record does not assert independent reproduction.

### GPT OSS 20B

F1 Micro: 48.76. Result ID: 69ccd69747f34ce924b67d72.

[Reporting paper](<https://www.sota2.com/research/paper/69cca6cc0636e5bf4afcbe5c.md>)

Recorded evaluation configuration: {"Category":"Base LLM inference","Zero-shot":"true"}. This record does not assert independent reproduction.

### Qwen3 4B Thinking

F1 Micro: 48.22. Result ID: 69ccd69747f34ce924b67d6f.

[Reporting paper](<https://www.sota2.com/research/paper/69cca6cc0636e5bf4afcbe5c.md>)

Recorded evaluation configuration: {"Category":"Base LLM inference","Zero-shot":"true"}. This record does not assert independent reproduction.

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