Related Experiment Videos
Giuseppe Vella1, Francesca Sala2, Vincenzo Pisciotta2
1Università di Palermo - Azienda sanitaria provinciale di Palermo.
Recenti Progressi in Medicina
|October 2, 2025
Summary
We developed an AI pipeline to automate epidemiological investigations, reducing investigation time by 60%. Expert validation confirmed high scores for completeness, accuracy, and relevance, demonstrating a scalable solution for public health surveillance.
Area of Science:
- Public Health
- Epidemiology
- Artificial Intelligence
Background:
- Epidemiological investigations are crucial for disease surveillance.
- Manual data extraction and report generation are time-consuming.
- Automation can improve efficiency and timeliness in public health.
Purpose of the Study:
- To implement and validate a three-phase AI pipeline for automating epidemiological investigations.
- To assess the performance of the AI system against expert standards.
- To evaluate the potential for time savings and scalability.
Main Methods:
- A three-phase AI pipeline was developed, including structured PDF data extraction, Retrieval-Augmented Generation (RAG)-driven report generation, and document assembly.
- Expert validation was conducted with 200 participants.
- Performance metrics included completeness, accuracy, relevance, clarity, and timeliness.
Main Results:
- The AI pipeline achieved high expert validation scores: completeness (4.7), accuracy (4.5), relevance (4.6), clarity (4.8), and timeliness (4.4).
- High inter-rater reliability was observed (κ=0.85).
- A 60% reduction in investigation time was achieved.
Conclusions:
- The AI pipeline effectively automates epidemiological investigations.
- The system demonstrates high performance and significant time savings.
- The scalable nature of the system allows for application in various public health surveillance contexts.