Swarm Learning for Decentralised and Confidential Clinical Machine Learning

One of the main goals of precision medicine is to identify patients with severe and heterogeneous illnesses.

There are patients with leukaemia being identified with basic machine learning using their blood transcriptomes.

However, there are complications when it comes to determining what is allowed and what is possible due to the current privacy legislation.

This is why Swarm Learning has been developed.

It is working on the integration of medical data from data owners without violation of privacy laws.

By using cutting-edge blockchain technology and decentralised machine learning there is no need for a central coordinator, and therefore goes beyond federated learning.

To display the practicability of Swarm Learning and how it develops its disease classifiers, it was given 4 heterogeneous diseases: COVID-19, tuberculosis, leukaemia, and lung pathologies.

A recent paper took more than 16,400 blood tanscriptomes from 127 clinical studies alongside 95,000 chest x-rays.

The paper found that Swarm learning classifiers outperformed those developed at individual sites.

It is thought that Swarm Learning could noticeably accelerate the introduction of precision medicine.

Source: Nature