When blockchain meets artificial intelligence: An application to cancer histopathology

Area of Science:

• Oncology
• Computer Science
• Bioinformatics

Background:

• Accurate prediction of molecular features from histopathology images is crucial for cancer diagnosis and treatment.
• Current predictive models often rely on centralized data, posing challenges for data privacy and accessibility.

Purpose of the Study:

• To evaluate the efficacy of blockchain-based models compared to traditional local models for predicting molecular features in cancer histopathology.
• To assess the performance of decentralized machine learning models in the context of cancer molecular diagnosis.

Main Methods:

• Utilized blockchain technology to develop decentralized predictive models.
• Compared the performance of blockchain-based models against local and merged models using cancer histopathology images.
• Focused on predicting specific molecular features relevant to cancer.

Main Results:

• Blockchain-based models demonstrated superior performance compared to local models.
• Decentralized models achieved performance comparable to merged models, indicating robustness and scalability.
• The study confirmed the potential of blockchain in enhancing molecular diagnosis accuracy.

Conclusions:

• Blockchain-based decentralized models are a viable and effective alternative for predicting molecular features from histopathology images.
• The findings support the integration of blockchain technology into cancer diagnostics for improved accuracy and data security.
• Decentralized approaches offer a promising future for molecular cancer diagnosis.