Nat Med uses artificial intelligence to safely detect cancer from patient data

Artificial intelligence (AI) can analyze large amounts of data, such as images or test results, and identify patterns that are often undetectable by humans, making it extremely valuable in accelerating disease detection, diagnosis and treatment

However, the use of the technology in a healthcare setting is controversial because of the risk of accidental data breaches and the fact that many systems are owned and controlled by private companies that have access to confidential patient data and a responsibility to protect it

Researchers are beginning to investigate whether a type of artificial intelligence known as swarm learning can help computers predict cancer in medical images of patient tissue samples without releasing hospital data

Swarm learning trains artificial intelligence algorithms to detect patterns in local hospital or university data, such as genetic changes in images of human tissue

By doing this multiple times, the algorithm can be improved and one can be created that works on all datasets

The team trained the AI ​​algorithm on study data from three groups of patients in Northern Ireland, Germany and the United States

The research was led by Jakob Nicholas Kaiser, Visiting Associate Professor at the University of Leeds School of Medicine and Research Fellow at RWTH Aachen University Hospital

"Based on data from more than 5,000 patients, we were able to demonstrate that an AI model trained with swarm learning could predict clinically relevant genetic changes directly from images of colon tumor tissue," said Dr.

Phil Quark, Professor of Pathology at the University of Leeds School of Medicine, said: "We have shown that, in medicine, swarm learning can be used to train stand-alone AI algorithms for any image analysis task

"Creating an AI system that can accomplish this task could improve our ability to apply AI in the future

Journal Reference :

1. Saldanha, OL, Quirke, P.