Initial commit

config/edsnlp_config.yaml

@@ -0,0 +1,46 @@
+# EDS-NLP Configuration
+# Configuration for EDS-NLP integration in the MCO PMSI coding pipeline
+
+# Model configuration
+model_name: "fr_core_news_sm"  # spaCy French model
+
+# Component toggles - enable/disable specific EDS-NLP components
+enable_sentences: true
+enable_negation: true
+enable_hypothesis: true
+enable_history: true
+enable_family: true
+enable_reported_speech: true
+
+# Performance tuning
+cache_pipeline: true
+batch_size: 32
+max_length: 1000000  # Maximum document length in characters
+
+# Fallback configuration
+enable_fallback: true
+max_failures_before_cooldown: 3
+cooldown_period_seconds: 300  # 5 minutes
+
+# Timeout configuration
+processing_timeout_seconds: 30.0
+
+# Normalization
+enable_normalization: true
+abbreviations_file: "config/medical_abbreviations.json"
+
+# Entity extraction configuration
+extract_diagnostics: true
+extract_medications: true
+extract_procedures: true
+extract_dates: true
+extract_measurements: true
+
+# Confidence thresholds
+min_entity_confidence: 0.3
+min_qualifier_confidence: 0.5
+
+# Logging configuration
+log_processing_time: true
+log_entity_counts: true
+performance_warning_threshold: 1.0  # seconds