EUVIMEDEUVIMED entdecken
VergleichsstudieOpen Access · CC BY 4.0

The Artificial Intelligence-Assisted Diagnosis of Skeletal Dysplasias in Pediatric Patients: A Comparative Benchmark Study of Large Language Models and a Clinical Expert Group

Nikola Ilić iD, Nina Marić iD, Dimitrije Cvetković, Marko Bogosavljević, Gordana Bukara-Radujković, Jovana Krstić, Zoran Paunović, Ninoslav Begović iD, Sanja Panić Zarić, Slađana Todorović, Katarina Mitrović iD, Aleksandar Vlahović und 1 weitere

Genes · 2025

DOI: 10.3390/genes16070762 ↗

Kurz gefasst

Vergleichsstudie in der Zeitschrift Genes, 2025. Verfasst von Nikola Ilić und 12 weiteren Autor:innen. Beteiligte Einrichtung: Clinical Genetics Outpatient Clinic, Mother and Child Health Care Institute of Serbia "Dr Vukan Cupic", 11070 Belgrade, Serbia. Hauptthemen laut MEDLINE: Artificial Intelligence, Bone Diseases, Developmental. Die Arbeit wurde bisher 11-mal zitiert (Crossref) und nennt 46 Literaturangaben. Der Volltext ist frei zugänglich (Open Access, CC BY 4.0).

AbstractSchlagwörterBibliografische AngabenZitierenVerwandte Studien

Abstract (Originalsprache)

Background/Objectives: Skeletal dysplasias are a heterogeneous group of rare genetic disorders with diverse and overlapping clinical presentations, posing diagnostic challenges even for experienced clinicians. With the increasing availability of artificial intelligence (AI) in healthcare, large language models (LLMs) offer a novel opportunity to assist in rare disease diagnostics. This study aimed to compare the diagnostic accuracy of two advanced LLMs, ChatGPT (version GPT-4) and DeepSeek, with that of a clinical expert panel in a cohort of pediatric patients with genetically confirmed skeletal dysplasias.

Methods

We designed a prospective vignette-based diagnostic benchmarking study including 45 children with confirmed skeletal dysplasias from two tertiary centers. Both LLMs were prompted to provide primary and differential diagnoses based on standardized clinical case vignettes. Their outputs were compared with those of two human experts (a pediatric endocrinologist and a pediatric orthopedic surgeon), using molecular diagnosis as the gold standard.

Results

ChatGPT and DeepSeek achieved a comparable diagnostic top-3 accuracy (62.2% and 64.4%, respectively), with a high intermodel agreement (Cohen’s κ = 0.95). The expert panel outperformed both models (82.2%). While LLMs performed well on more common disorders, they struggled with ultra-rare and multisystemic conditions. In one complex case missed by experts, the DeepSeek model successfully proposed the correct diagnosis.

Conclusions

LLMs offer a complementary diagnostic value in skeletal dysplasias, especially in under-resourced medical settings. Their integration as a supportive tool in multidisciplinary diagnostic workflows may enhance early recognition and reduce diagnostic delays in rare disease care.

Schlagwörter

MeSH-Schlagwörter (MEDLINE)

Artificial IntelligenceBone Diseases, DevelopmentalHumansChildMaleFemaleChild, PreschoolBenchmarkingProspective StudiesInfantAdolescentLarge Language Models

Stichwörter der Autor:innen

ChatGPTDeepSeekartificial intelligence (AI)diagnostic benchmarkinglarge language models (LLMs)pediatric geneticsskeletal dysplasia

Autor:innen und Institutionen

  • Clinical Genetics Outpatient Clinic, Mother and Child Health Care Institute of Serbia "Dr Vukan Cupic", 11070 Belgrade, Serbia.
  • Clinic for Children Diseases, University Clinical Center of the Republic of Srpska, 78000 Banja Luka, Bosnia and Herzegovina.
  • Medical Faculty, University of Banjaluka, 78000 Banja Luka, Bosnia and Herzegovina.
  • Department of Endocrinology, Mother and Child Health Care Institute of Serbia "Dr Vukan Cupic", 11070 Belgrade, Serbia.
  • Department of Orthopedic Surgery and Traumatology, Mother and Child Health Care Institute of Serbia "Dr Vukan Cupic", 11070 Belgrade, Serbia.
  • Department of Surgery, Division of Pediatric Surgery, University of Belgrade, 11000 Belgrade, Serbia.

ORCID-Profile: Nikola Ilić, Nina Marić, Ninoslav Begović, Katarina Mitrović, Adrijan Sarajlija

Zitieren

APA

Ilić, N., Marić, N., Cvetković, D., Bogosavljević, M., Bukara-Radujković, G., Krstić, J., Paunović, Z., Begović, N., Panić Zarić, S., Todorović, S., Mitrović, K., Vlahović, A., & Sarajlija, A. (2025). The Artificial Intelligence-Assisted Diagnosis of Skeletal Dysplasias in Pediatric Patients: A Comparative Benchmark Study of Large Language Models and a Clinical Expert Group. Genes. https://doi.org/10.3390/genes16070762

Vancouver

Ilić N, Marić N, Cvetković D, Bogosavljević M, Bukara-Radujković G, Krstić J, et al. The Artificial Intelligence-Assisted Diagnosis of Skeletal Dysplasias in Pediatric Patients: A Comparative Benchmark Study of Large Language Models and a Clinical Expert Group. Genes. 2025. doi:10.3390/genes16070762

BibTeX
@article{Ili2025,
  author = {Ilić, Nikola and Marić, Nina and Cvetković, Dimitrije and Bogosavljević, Marko and Bukara-Radujković, Gordana and Krstić, Jovana and Paunović, Zoran and Begović, Ninoslav and Panić Zarić, Sanja and Todorović, Slađana and Mitrović, Katarina and Vlahović, Aleksandar and Sarajlija, Adrijan},
  title = {The Artificial Intelligence-Assisted Diagnosis of Skeletal Dysplasias in Pediatric Patients: A Comparative Benchmark Study of Large Language Models and a Clinical Expert Group},
  journal = {Genes},
  year = {2025},
  doi = {10.3390/genes16070762}
}
Kostenlos starten →