You can deploy this model and start earning money today!

raphael-r/
bright-gliner-ihc

Name: bright-gliner-ihc
Availability: InStock

Views

No views yet

BRIGHT NER: GLiNER2 fine-tuned for ihc

Description

This is a GLiNER2 architecture fine-tuned to extract clinical neuro-oncology entities related to the ihc semantic group. It was trained on a synthetic dataset generated for the properly de-identified BRIGHT project dataset (see the generated_data folder in the primary repository).

This model repository was specifically designed to fit within the bright_db overarching namespace.

Fields

It extracts the following fields (described in French):

ihc_idh1: Expression IDH1 R132H (positif/négatif)
ihc_atrx: ATRX (conservé/perdu)
ihc_p53: Expression p53
ihc_fgfr3: Expression FGFR3
ihc_braf: Expression BRAF V600E
ihc_gfap: GFAP (positif/négatif)
ihc_olig2: Olig2
ihc_ki67: Index Ki-67 (0-100%)
ihc_hist_h3k27m: Expression H3K27M
ihc_hist_h3k27me3: Expression H3K27me3
ihc_egfr_hirsch: EGFR score de Hirsch
ihc_mmr: Protéines réparation mésappariements

Performance on Validation Set

Aggregates:

Macro F1: 0.4494 (Precision: 0.3754, Recall: 0.8047)
Micro F1: 0.6059 (Precision: 0.4412, Recall: 0.9669)

Per-Label Breakdowns:

Label	Precision	Recall	F1
ihc_idh1	0.7085	0.9860	0.8246
ihc_atrx	0.7069	0.9389	0.8066
ihc_p53	0.6324	0.9435	0.7573
ihc_fgfr3	0.0000	0.0000	0.0000
ihc_braf	0.0714	1.0000	0.1333
ihc_gfap	0.7605	0.9922	0.8610
ihc_olig2	0.7750	0.9841	0.8671
ihc_ki67	0.6477	0.9542	0.7716
ihc_hist_h3k27m	0.1250	1.0000	0.2222
ihc_hist_h3k27me3	0.0390	1.0000	0.0750
ihc_egfr_hirsch	0.0385	0.8571	0.0736
ihc_mmr	0.0000	0.0000	0.0000

Usage

python

# Inference Code
from gliner2 import GLiNER2

model = GLiNER2.from_pretrained("raphael-r/bright-gliner-ihc")
text = "Patient presenting with epileptic seizures..."
entities = model.extract_entities(text)

for entity in entities:
    print(entity["text"], "=>", entity["label"])