Scientists have used artificial intelligence to improve prediction models used for brain cancer metastases in people.

The researchers used a technique called a paste-targeted AI to create a machine-learning framework for using PET scan images to predict patients age.

The team reported that the model outperformed almost two out of three cryo-electron microscopy (cryo-EM) nanopore computed tomography (NEM) nanopore detectors and the network-based prediction was superior to the traditional post-mortem tissue autopsy model.

Dr. Alex Cempe Cancer Research UKs Chief Scientist in Cancer Research said: Alzheimers disease is the most common cause of dementia and around one in eight adults are affected by this debilitating and potentially fatal disease. That means that Alzheimers and related dementias are increasingly difficult to treat-with the addition of powerful new drugs returning them to health.

However current imaging technologies are relatively new and under-tested taking up to 6 years to achieve a single measurement of Alzheimers biomarkers. In addition to costs and standardization of the equipment current methods to test patients and evaluate model performance are manual and expensive and may cause long-term risks to patients.

These problems are compounded when more obscure biomarkers such as small molecules may not have yet been detected. For example myeloid markers may not yet be fully detectable in the brain due to acquired hippocampal changes.

Therefore making detection of the existing biomarkers more accurate and more quickly increasing the rate of detection in real time is extremely important. This is as the case for all disease biomarkers.

A holistic approach to developing tests and development techniques to detect Alzheimers biomarkers is therefore essential in achieving practical and routine testing. This includes the use of novel affordable machine-learning based tests to help drive the development of novel approaches for detection of biomarkers (such as PET scan) in the brain.

Professor Roger Watson Professor of Computer Science at the University of Surrey said: Although the use of AI to improve neurological biomarker (biomarkers) has been pioneered by our study AI actually offers potential to do more here than in other applications in medicine.

Specifically our study allows us to improve upon current diagnostic methods in a range of innovative ways including providing a means for the identification of disease biomarkers that have grown into huge volumes within biological databases.

Other researchers involved in the study include Dr. Tim Raine (Lancet) and Dr. Richard Taylor (British Columbia Health Sciences Centre Vancouver Canada) and Professor Keith Walker (Lancet).