EUVIMEDEUVIMED entdecken
Open Access · CC BY-NC-ND 4.0

A Locally Executed Agentic Artificial Intelligence Framework for Deduplication, Screening, and Structured Data Extraction in Spine Surgery Systematic Reviews

Sathish Muthu iD, Dhibin Vikash Kolarpatti Ponnusamy iD, Vibhu Krishnan Viswanathan iD, Sathish Kumar Rajappan Chandra iD, Khan Sharun

JBJS Open Access · 2026-07 · 11 (3)

DOI: 10.2106/jbjs.oa.26.00102 ↗

Kurz gefasst

Fachartikel in der Zeitschrift JBJS Open Access, 2026-07. Verfasst von Sathish Muthu und 4 weiteren Autor:innen. Beteiligte Einrichtung: Department of Spine Surgery, Orthopaedic Research Group, Coimbatore, Tamil Nadu, India. Die Arbeit nennt 46 Literaturangaben. Der Volltext ist frei zugänglich (Open Access, CC BY-NC-ND 4.0).

AbstractBibliografische AngabenZitierenVerwandte Studien

Abstract (Originalsprache)

Systematic reviews and meta-analyses remain time-consuming and labor-intensive. We developed and validated a locally executed agentic artificial intelligence (AI) framework for deduplication, screening, and structured data extraction in systematic reviews, reported following Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA)-Transparent Reporting of AI in Comprehensive Evidence Synthesis (trAIce) guidelines. A multiagent pipeline of specialized agents for deduplication, title/abstract screening, structured data extraction, and verification was executed entirely locally to ensure data governance and reproducibility. Human-in-the-loop validation compared outputs against dual independent reviewers across 3 spine surgery systematic reviews, assessing accuracy, inter-rater agreement (Cohen’s κ), time savings, and clinically critical error rates. Across 6,214 records, deduplication achieved near-perfect agreement with human reviewers (κ = 0.98), title and abstract screening yielded higher concordance than human screening (κ = 0.91) while reducing full-text review volume by 83%, and structured data extraction reached substantial agreement (κ = 0.87). The framework reduced reviewer time by 91.1% (90.8%-91.4%), a mean saving of 19.5 hours per review (p < 0.001). Clinically critical discrepancies were rare (

Autor:innen und Institutionen

  • Department of Spine Surgery, Orthopaedic Research Group, Coimbatore, Tamil Nadu, India.
  • Central Research Laboratory, Aarupadai Veedu Medical College and Hospital, Vinayaka Mission's Research Foundation (Deemed to be University), Puducherry, India.
  • Department of Orthopaedics, Government Medical College, Karur, Tamil Nadu, India.
  • Department of Orthopaedic Surgery, University of Alabama, Birmingham, Alabama.
  • Clinical Trial & Research Unit, Interdisciplinary Institute of Indian System of Medicine (IIISM), SRM Institute of Science and Technology, Kattankulathur, Tamil Nadu, India.
  • Graduate Institute of Medicine, Yuan Ze University, Taoyuan, Taiwan.

ORCID-Profile: Sathish Muthu, Dhibin Vikash Kolarpatti Ponnusamy, Vibhu Krishnan Viswanathan, Sathish Kumar Rajappan Chandra

Zitieren

APA

Muthu, S., Kolarpatti Ponnusamy, D. V., Viswanathan, V. K., Rajappan Chandra, S. K., & Sharun, K. (2026). A Locally Executed Agentic Artificial Intelligence Framework for Deduplication, Screening, and Structured Data Extraction in Spine Surgery Systematic Reviews. JBJS Open Access, 11(3). https://doi.org/10.2106/jbjs.oa.26.00102

Vancouver

Muthu S, Kolarpatti Ponnusamy DV, Viswanathan VK, Rajappan Chandra SK, Sharun K. A Locally Executed Agentic Artificial Intelligence Framework for Deduplication, Screening, and Structured Data Extraction in Spine Surgery Systematic Reviews. JBJS Open Access. 2026;11(3). doi:10.2106/jbjs.oa.26.00102

BibTeX
@article{Muthu2026,
  author = {Muthu, Sathish and Kolarpatti Ponnusamy, Dhibin Vikash and Viswanathan, Vibhu Krishnan and Rajappan Chandra, Sathish Kumar and Sharun, Khan},
  title = {A Locally Executed Agentic Artificial Intelligence Framework for Deduplication, Screening, and Structured Data Extraction in Spine Surgery Systematic Reviews},
  journal = {JBJS Open Access},
  year = {2026},
  volume = {11},
  number = {3},
  doi = {10.2106/jbjs.oa.26.00102}
}
Kostenlos starten →