Resource-Conscious Modeling for Next- Day Discharge Prediction Using Clinical Notes

1. 1Collect and preprocess postoperative clinical notes
2. 2Fine-tune compact large language models (LLMs) such as DistilGPT-2 and Bio_ClinicalBERT
3. 3Compare performance of LLMs with traditional text-based models like TF-IDF with XGBoost and LGBM
4. 4Evaluate models based on prediction accuracy and resource efficiency

Data scientists and AI engineers on a healthcare team can benefit from this research to improve resource allocation and bed turnover in elective spine surgery units

💡 Compact LLMs can be effective for predicting next-day discharge with minimal computational resources

📊 Predicting next-day discharge with lightweight models and clinical notes 💡