ChatGPT for digital pathology research

Mohamed Omar ¹, Varun Ullanat ², Massimo Loda ³

⁰Department of Pathology and Laboratory Medicine, Weill Cornell Medicine, New York, NY, USA.

The Lancet. Digital Health +

|

July 10, 2024

View abstract on PubMed

Summary

Domain-specific artificial intelligence (AI) tools, like those in digital pathology, improve medical research accuracy. Tailored large language models (LLMs) enhance information retrieval and democratize computational pathology access.

Area Of Science

Artificial Intelligence in Medicine
Digital Pathology
Computational Pathology

Background

Generative artificial intelligence (AI) models, such as ChatGPT, offer new possibilities for medical research.
Large language models (LLMs) require domain-specific adaptation for complex fields like digital pathology.
Existing LLMs have limitations in specialized medical domains, necessitating tailored solutions.

Purpose Of The Study

To explore the integration and challenges of LLMs in digital pathology.
To highlight the potential of domain-specific AI tools for accurate information retrieval.
To discuss the broader implications of AI in streamlining scientific research and computational pathology access.

Main Methods

Development of a domain-specific AI tool for digital pathology.
Utilizing a curated literature database and a user-interactive web application.
Focusing on tailored AI approaches to minimize inaccurate responses.

Main Results

Domain-specific AI tools enhance the reliability and accuracy of information extraction in digital pathology.
Curated databases and interactive applications facilitate precise, referenced information retrieval.
Tailored LLMs reduce the risk of erroneous outputs in specialized medical contexts.

Conclusions

Domain-specific AI tools are crucial for advancing digital pathology and medical research.
These tools can democratize access to computational pathology for researchers with limited coding expertise.
Integration of domain-specific AI in academia supports continuous learning and adaptation in medical research.