AI May Predict Cancer Without Putting Patient Data at Risk

A team of researchers recently developed a new method of using artificial intelligence or AI to forecast cancer from patient data minus putting personal information at risk.

As indicated in a News-Medical.net report, this new technology by the team which includes University of Leeds medical scientists can assess large amounts of data like images or test results and can determine patterns frequently undetectable by humans, making it highly valuable in fast-tracking the detection, diagnosis, and treatment of disease.

Nevertheless, with the use of AI in medical settings is controversial due to the risk of accidental data release, and many systems are owned and regulated by private companies, giving them access to confidential information of patients, as well as the responsibility that protects it.

The team set out to find out of a form of artificial intelligence, known as "swarm learning" could be used to help computers predict cancer in patient tissue samples' medical images, minus the need of releasing data from hospitals.

Swarm Learning

Essentially, swarm learning trains AI algorithms to identify patterns in data in a local university or hospital as genetic modifications within images of human tissue.

Then, the warm learning system is sending this newly trained algorithm, although essentially, no local data or patient information, to a central computer.

In that particular location, it is combined with algorithms identically yielded by other hospitals to develop an optimized algorithm.

This then is sent back to the local hospital where it is reapplied to the original data, enhancing the detection of genetic modifications due to its more sensitive detection abilities.

Bay doing this several times, the algorithm can be enhanced and one created that is working on all data sets.

Meaning, the approach can be used minus the need for any data to be provided to third-party companies, or be sent between hospitals throughout borders.

Cancer Prediction

In their study published in the Nature Medicine journal, the researchers trained algorithms on research data from three groups of patients from the United States, Germany, and Ireland.

Such algorithms were tested on two huge sets of data images produced at Leeds and were found to have learned successfully, how the existence of different subtypes of cancer in the images can be predicted.

This particular study was led by Associate Professor Jakob Nikolas Kather from the University of Leeds' School of Medicine, and a researcher at the University Hospital RWTH Aachen.

The research team includes Heike Grabsch, Dr. Nick West, and Dr. Phil Quirke, from the University of Leeds' School of Medicine.

A similar Medical Xpress report said that according to Dr. Kather, based on data from more than 5,000 patients, they were able to show that artificial intelligence models trained with swarm learning can clinically predict relevant genetic modifications directly from images of tissue from colon tumors.

Related information about AI being able to diagnose disease is shown on TED's YouTube video below:

Check out more news and information on Artificial Intelligence and Cancer in Science Times.