Nordic Life Science 1
EDTECHLABS, THE PATIENTcentered research centre r
un by KTH Royal Institute of Technology (KTH), Karolinska Institutet (KI) and Region Stockholm, has adopted two new research programs for 2020. When the centre opened a call for a second research area the requirement was that the results could benefit healthcare already within five years. The new programs are therefore based on established research. The programs will focus on AI and bioelectronic medicine, for application in the areas of breast cancer and inflammatory disease, respectively. “The research programs at MedTechLabs are, in each case, led by two scientists, one from KI and one from KTH, as it is this combination of unique expertise and novel ideas that we want to nurture,” says Peta Sjölander, Director of MedTechLabs. The first new program uses AI and machine learning to radically increase the accuracy of breast cancer imaging diagnostics. The research will be led by Associate Professor and clinician Johan Hartman, researcher at KI, and Associate Professor Kevin Smith, researcher at KTH and SciLifeLab. The researchers will use data from all patients diagnosed with breast cancer through mammography in the Stockholm region during the period 2005 and 2019. “This important research program is possible only through Sweden’s unique access to comprehensive and quality-assured patient data,” says Peta Sjölander. The research project is divided into two parts and will last for five years – although the aim is that it will last longer. “The first part is Radiology and by training the AI models using a great number of mammography images we will be able to help radiologists to detect cancer faster and more efficiently. In addition, we would like to be able to predict which patients will get cancer based on mammography screening images. The second part is Pathology where we are building the world’s largest database of microscopy images with appurtenant clinical data. This is being done by scanning microscopy glasses from the Stockholm region’s archive and from hospitals and connecting data from quality registries. The material is used to train AI models to detect cancer in microscopy images and predict clinically important parameters, such as risk for relapse or treatment resistance. All in all, we are investigating different AI models that may work as a future decision basis for doctors making diagnoses. These methods will contribute to more individually adjusted diagnostics or what is known as precision medicine,” says Johan Hartman. To establish a routine within healthcare, however, he adds that extensive testing, validation and regulatory approvals are required. “Therefore we are hoping to have a close collaboration 38 NORDICLIFESCIENCE.ORG