Book chapter
Towards Multidisciplinary Summarization of Hospital Stays: Efficient Sentence-Level Clinical Section Categorization
Artificial Intelligence in Medicine, pp.383-388
Lecture Notes in Computer Science, v. 16749, Springer Nature Switzerland
2026
DOI: 10.1007/978-3-032-30813-9_71
Abstract
Effective “all-team” summarization in high-complexity settings like the Neonatal Intensive Care Unit (NICU) requires aggregating insights from diverse disciplines (physicians, nurses, therapists) spread across hundreds of clinical free-text notes. Simply pooling heterogeneous text often leads to incoherent outputs. Structured summarization therefore first requires accurate categorization of sentence-level section labels across multi-source notes. This pilot study introduces a clinical section categorization pipeline using supervised fine-tuning (SFT) of large language models (LLMs). We adapted two Llama-3 models (8B and 70B) to MedSecId, a corpus of 2,002 MIMIC-III (Adult ICU) notes annotated with clinical section headers, achieving in-domain Macro F1 scores above 92% for both models. To evaluate cross-domain generalization, we assessed model capacity (8B vs. 70B) and quantization on a gold-standard dataset of 227 sentence-level spans derived from three multidisciplinary NICU summaries. Experimental results suggest a potential scale-dependent transfer effect: while SFT produced only marginal changes for the 8B model, it substantially improved the 70B model, increasing Macro F1 by 7%. Notably, the quantized fine-tuned 70B model outperformed its full-precision baseline while substantially reducing computational requirements. These findings suggest that sufficient model capacity is critical for preserving semantic flexibility during cross-domain clinical transfer and that efficient quantized adaptation can enable structured section-label modeling for downstream summarization.
Details
- Title: Subtitle
- Towards Multidisciplinary Summarization of Hospital Stays: Efficient Sentence-Level Clinical Section Categorization
- Creators
- Baris Karacan - University of Illinois ChicagoVaibhav Bhargava - University of Illinois ChicagoBarbara Di Eugenio - University of Illinois ChicagoNatalie Parde - University of Illinois ChicagoMary Khetani - University of Illinois ChicagoYu-Shan TsengVanessa Barbosa - University of Illinois ChicagoJulie Vignato - University of IowaLindsey Knake - University of IowaRajashree Dahal - University of Illinois ChicagoEmily Spellman - University of IowaDanielle Hitzel - University of IowaJanine Petitgout - University of IowaKristi Haughey - University of IowaAmanda Karstens - University of IowaBrianna Clarahan - University of IowaRachel Dawson - University of IowaLauren BoydMackenzie Weis - University of MissouriAngie Tipton - University of MissouriJaewon Bae - University of Iowa, NursingCatherine K. Craven - University of MissouriKaren Dunn Lopez - University of IowaAndrew D. Boyd - University of Illinois Chicago
- Contributors
- Pavel Andreev (Editor)William Van Woensel (Editor)John Holmes (Editor)Antoine Sauré (Editor)
- Resource Type
- Book chapter
- Publication Details
- Artificial Intelligence in Medicine, pp.383-388
- Series
- Lecture Notes in Computer Science; v. 16749
- DOI
- 10.1007/978-3-032-30813-9_71
- eISSN
- 1611-3349
- ISSN
- 0302-9743
- Publisher
- Springer Nature Switzerland; Cham
- Language
- English
- Date published
- 2026
- Academic Unit
- Stead Family Department of Pediatrics; Nursing; Neonatology; Otolaryngology
- Record Identifier
- 9985180785602771
Metrics
1 Record Views