Artificial Intelligence (AI) for health and datadriven decision-making

What is AI and deep learning?

AI is a collective term for systems or machines that mimic our intelligence. This allows them to perform tasks for us, and improve themselves at the same time.

Part of AI is machine learning: this involves a learning system that can improve itself based on the data it receives. For example, chatbots that understand your problems and automatically provide an answer, or a program that gives you automatic recommendations for movies based on your viewing habits.

Deep learning

A variant of machine learning is deep learning. An AI system can recognize complex patterns in large data sets by using algorithms that mimic the workings of the human brain, so-called 'deep neural networks'. The difference between deep learning and the broader term 'machine learning' is the accuracy of analyses the algorithms do: they are so thorough that they are similar to the processes in our brains. A network of these algorithms can therefore process large amounts of data and learn from it.

Data is everywhere in hospitals: patient records, clinical measurements, genetic information, text reports, and images. That data forms the basis of diagnoses, decisions, treatments, monitoring and real-time support during operations and procedures, for example.

Support in diagnostics

It is possible to recognize patterns in that medical data, for example, to aid in diagnostics. In other medical fields where large amounts of data are available, such as genetics, temporal monitoring or large texts (patient records), deep learning can also improve scientific research, and thus clinical practice.

AI can support clinical decisions and automate tasks, thereby reducing costs and keeping healthcare affordable. It can also improve healthcare by finding better ways to extract clinically useful information from data.

How does AI help our patient and healthcare?

AI can add value to healthcare and to patients in several ways.

For starters, it can help people at home gain more insight into their health, and based on that, get them the right care faster. One example is the WondGezond app (in Dutch), which is in development at Radboudumc. Someone recovering from surgery at home can use this app to take a picture of their wound. The app will then use AI to analyze the wound to determine if it is healing properly. If everything looks good, the app will let you know immediately. If in doubt, a doctor can review the photo and, if necessary, tailored care is available.

Diagnostics using AI

AI can also help healthcare providers in making diagnoses, for example when analyzing radiological or pathological images. In oncological hospital care, this is already being applied or further developed in various ways, for example by Radboudumc spin-offs Aiosyn and Thirona. But also in the practice of the general practitioner or obstetrician, more and more opportunities for AI are emerging. For example, Ardim - another Radboudumc spin-off - is working on an app that makes it possible to use a cell phone to diagnose hip dysplasia in unborn children. Delft Imaging is using similar technology in its BabyChecker app, which allows health care providers in Africa and Southeast Asia to detect potential problems early during pregnancy. This AI technology was also co-developed at Radboudumc.

Radboudumc spin-offs in AI

• Aiosyn

  AI application for automated quality control of pathological images.

  more on Aiosyn

• Ardim

  Smartphone AI applications for medical imaging 'at the point of care'.

  more on Ardim

• Screenpoint Medical

  AI applications for breast cancer detection.

  more on Screenpoint Medical

• Thirona

  AI applications for analysis of chest CT scans.

  more on Thirona