---
license: apache-2.0
language:
- zh
tags:
- medical
- perioperative
- complications
- lora
- adapter
- clinical-ai
datasets:
- perioperative-complications
pipeline_tag: text-classification
---

# PeriComp: Perioperative Complication Detection LoRA Adaptors

![PeriComp Performance](../figure6b.png)
*Figure: Performance comparison of fine-tuned models across different sizes*

## 🩺 Model Overview

**PeriComp** is a collection of specialized LoRA (Low-Rank Adaptation) adaptors designed for **perioperative complication detection** from clinical narratives. These adaptors enhance smaller open-source language models to achieve expert-level performance in identifying and grading 22 distinct perioperative complications based on European Perioperative Clinical Outcome (EPCO) definitions.

### 🎯 Key Features

- **Expert-level Performance**: Matches or exceeds human clinician accuracy
- **Multi-scale Detection**: Simultaneous identification and severity grading (mild/moderate/severe)
- **Comprehensive Coverage**: 22 distinct perioperative complications
- **Resource Efficient**: Optimized for deployment on standard clinical infrastructure
- **Privacy Preserving**: Fully deployable on-premises without data transmission

## 📊 Model Collection

This collection includes five optimized LoRA adaptors:

| Model | Base Model | Parameters | F1 Score | Use Case |
|-------|------------|------------|----------|----------|
| **PeriComp-4B** | Qwen3-4B | 4B | 0.55 | Resource-constrained environments |
| **PeriComp-8B** | Qwen3-8B | 8B | 0.61 | Balanced performance/efficiency |
| **PeriComp-14B** | Qwen3-14B | 14B | 0.65 | High-performance deployment |
| **PeriComp-32B** | Qwen3-32B | 32B | 0.68 | Maximum accuracy requirements |
| **PeriComp-QwQ-32B** | QwQ-32B | 32B | 0.70 | Reasoning-enhanced performance |

## 🔬 Research Background

Perioperative complications affect millions of patients globally, with traditional manual detection suffering from:
- **27% under-reporting rate** in clinical registries
- **High variability** in expert performance across institutions
- **Cognitive load limitations** with complex documentation

Our research demonstrates that targeted task decomposition combined with LoRA fine-tuning enables smaller models to achieve expert-level diagnostic capabilities while maintaining practical deployability.

![Strict Performance Evaluation](../figure7.png)
*Figure: Strict performance evaluation requiring exact complication type and severity matching*

## 🚀 Quick Start

### Installation

```bash
pip install transformers peft torch
```

### Basic Usage

```python
from transformers import AutoTokenizer, AutoModelForCausalLM
from peft import PeftModel

# Load base model and tokenizer
model_name = "Qwen/Qwen3-8B"
tokenizer = AutoTokenizer.from_pretrained(model_name)
base_model = AutoModelForCausalLM.from_pretrained(model_name)

# Load PeriComp adaptor
adaptor_name = "your-username/Qwen3-8B-PeriComp"
model = PeftModel.from_pretrained(base_model, adaptor_name)

# Prepare clinical input
clinical_text = '''
Patient Demographics: 65-year-old male
Procedure: Laparoscopic cholecystectomy
Postoperative Course: POD#2 - Patient reports abdominal pain, 
fever 38.5°C, elevated WBC count 15,000/μL...
'''

# Generate complication assessment
inputs = tokenizer(clinical_text, return_tensors="pt")
outputs = model.generate(**inputs, max_new_tokens=512)
result = tokenizer.decode(outputs[0], skip_special_tokens=True)
```

### Targeted Strategy Usage

For optimal performance with smaller models, use our targeted strategy:

```python
# Define complications to assess
complications = [
    "acute_kidney_injury",
    "surgical_site_infection", 
    "paralytic_ileus",
    # ... other complications
]

# Assess each complication individually
results = {}
for complication in complications:
    prompt = f"Assess for {complication}: {clinical_text}"
    # ... inference code
    results[complication] = assessment
```

## 📈 Performance Metrics

### Validation Results (Micro-averaged F1 Scores)

- **Center 1 (Primary)**: Complex tertiary care cases
- **Center 2 (External)**: Community hospital validation

| Model | Center 1 F1 | Center 2 F1 | Human Expert F1 |
|-------|-------------|-------------|-----------------|
| PeriComp-4B | 0.55 | 0.52 | 0.526 |
| PeriComp-8B | 0.61 | 0.58 | 0.526 |
| PeriComp-14B | 0.65 | 0.62 | 0.526 |
| PeriComp-32B | 0.68 | 0.65 | 0.526 |
| PeriComp-QwQ-32B | 0.70 | 0.67 | 0.526 |

### Key Advantages

✅ **Consistent Performance**: No degradation with document complexity  
✅ **24/7 Availability**: Continuous monitoring capability  
✅ **Standardized Assessment**: Eliminates inter-observer variability  
✅ **Comprehensive Detection**: All 22 EPCO-defined complications  
✅ **Privacy Compliant**: On-premises deployment option  

## 🔧 Technical Details

### Training Methodology

- **Base Architecture**: Qwen3 series and QwQ-32B
- **Fine-tuning Method**: LoRA (Low-Rank Adaptation)
- **Training Data**: 146 complex surgical cases
- **Validation**: Dual-center external validation (52 cases)
- **Task Strategy**: Targeted decomposition approach

### LoRA Configuration

```python
lora_config = {
    "lora_rank": 16,
    "lora_alpha": 32,
    "learning_rate": 1e-4,
    "target_modules": ["q_proj", "k_proj", "v_proj", "o_proj"]
}
```

## 📋 Supported Complications

The models detect and grade 22 perioperative complications:

1. **Cardiovascular**: Myocardial injury, cardiac arrhythmias
2. **Respiratory**: Pneumonia, respiratory failure  
3. **Renal**: Acute kidney injury
4. **Gastrointestinal**: Paralytic ileus, anastomotic leakage
5. **Infectious**: Surgical site infections, sepsis
6. **Neurological**: Delirium, stroke
7. **Hematological**: Bleeding, thromboembolism
8. **And more...**

Each complication is graded as:
- **Mild**: Minor intervention required
- **Moderate**: Significant medical management
- **Severe**: Life-threatening, intensive intervention

## 🏥 Clinical Applications

### Primary Use Cases

- **Automated Screening**: Continuous 24/7 complication monitoring
- **Quality Assurance**: Systematic complication registry validation  
- **Clinical Decision Support**: "Second opinion" for complex cases
- **Research**: Standardized outcome assessment for clinical studies

### Deployment Scenarios

- **Resource-limited Settings**: Use PeriComp-4B/8B models
- **Standard Clinical Environment**: PeriComp-14B recommended
- **High-accuracy Requirements**: PeriComp-32B for maximum performance
- **Reasoning-enhanced Tasks**: PeriComp-QwQ-32B for complex diagnostic reasoning

## ⚠️ Important Considerations

### Clinical Validation Required

⚠️ **These models are research tools and require clinical validation before use in patient care**

### Limitations

- Training on Chinese medical records (generalizability considerations)
- Performance depends on documentation quality and completeness
- Not a replacement for clinical judgment

### Best Practices

- Use as **screening tool** with clinical oversight
- Validate outputs against clinical judgment
- Consider local adaptation for specific institutional practices

## 📚 Citation

If you use PeriComp in your research, please cite:

```bibtex
@article{pericomp2025,
  title={Enhancing Local Language Models for Perioperative Complication Detection: A Targeted Strategy with LoRa Fine-tuning},
  author={[Authors]},
  journal={[Journal]},
  year={2025}
}
```

## 📧 Contact & Support

For questions, issues, or collaboration opportunities:

- **Research Team**: Department of Anesthesiology, First Affiliated Hospital of Sun Yat-sen University, Guangzhou, China
- **Technical Issues**: gaoshw5@mail.sysus.edu.cn
- **Clinical Applications**: Perioperative Complications Detection

## 📄 License

This work is licensed under Apache License 2.0. See LICENSE for details.

---

*PeriComp: Advancing perioperative patient safety through AI-powered complication detection*