A team of researchers from esteemed German research institutes have created an artificial intelligence (AI) medical companion that can diagnose patients and provide supporting evidence of its diagnoses.

While machine learning systems are increasingly being used to detect illnesses and conditions, the researchers felt that the conclusions reached by these systems are often incomprehensible to doctors and patients, making it difficult for them to accept the diagnosis without further testing.

“Machine learning processes help make diagnoses, but if their decisions are not comprehensible to doctors and patients, they have to be taken with a grain of salt and might even have to be ignored in sensitive fields like medicine,” Ute Schmid, Professor of Cognitive Systems at the University of Bamberg, said.

The “Transparent Medical Expert Companion” is made up of two systems. The first uses microscopy imaging techniques to look for signs of illness. The system is trained to recognise these signs thanks to a deep neural network that allows for the classification and analysis of large volumes of imaging data.

The second system looks within the deep learning network and uses the information to create supporting text and images that explains why the first system reached its decision. This is done using a process known as inductive logic programming, which is designed to provide rational to patients and medical professionals.

The system is currently only being tested to detect signs of colon cancer, but researchers hope to “develop companions for different medical experts that meet important criteria like transparency and explicability”.

The project is a joint collaboration between researchers from the The Fraunhofer Institute for Integrated Circuits IIS, the Fraunhofer Heinrich Hertz Institute HHI, the University of Erlangen’s Pathological Institute, and the University of Bamberg.

The project has been backed by the German Federal Ministry of Education and Research, which has pledged €1.3m in funding to the project until August 2021.

Improving diagnosis

We often hear how AI is set to take jobs away from human workers, but the Transparent Medical Expert Companion is designed to aid doctors, rather than replace them.

“The attending doctors decide whether or not they agree with the assessment,” University of Bamberg research assistant Bettina Finzel explained.

While the AI system may offer support in diagnosing a patient, the human doctor remains in control, and can make amendments and corrections within the system which will allow it to continue learning and improving over time.


Read more: How assistive technology is helping people with learning disabilities


 

Free download worth over $5000
Download our 2018 Technology, Media and Telecoms Predictions Report

Worth up to $5,850

In this report, we look at the 30 big tech themes for 2019, identifying winners and losers for each theme. This report will impact all industries helping:

  • CEOs/Senior Staff: in all industries understand the disruptive threats to their competitive landscape
  • Investors: Helps fund managers focus their time on the most interesting investment opportunities in global TM