Research project Predictive analysis in complex diseases offers opportunities that are crucial for identifying interventional targets at the individual, patient group, and healthcare management levels, thereby facilitating resource allocation for high-demand diseases.
This interdisciplinary project, including experts in machine learning, biostatistics, and stroke care, aims to develop high-performance, explainable machine and deep learning models that can accurately predict severe outcomes in patients with complex diseases, such as stroke. These models will leverage Sweden’s unique register data, integrate imaging, and generate new insights into optimal patient care.
The Swedish Research Council
Mia von Euler, professor of Neurology, Örebro University
András Büki, professor of Neurosurgery, Örebro University
Our project aims to develop explainable AI models that can identify patients and patient groups at the highest risk of severe outcomes with high precision and provide decision support for optimal care of complex diseases.
By using new AI-based methods (machine learning and deep learning), we can integrate different types of information sources and find complex patterns. This allows us to leverage the potential of Sweden’s unique health and medical registers, combined with, for example, imaging diagnostics, and create new knowledge on how to provide the best possible care. However, development is needed in terms of precision and interpretability to make the models practically applicable for decision support in healthcare.
Stroke (a clot or bleeding in the brain) is an example of a complex disease in terms of causes and treatment. In Sweden, approximately 23,000 cases occur annually. Three months after a stroke, a quarter of the patients are deceased or dependent on help. To achieve our overall goal, we will first develop and validate prediction models. Then, we will evaluate the models’ predictive ability and usefulness to understand the causes of a wide range of outcomes and identify effective interventions. Finally, we will model the effect of future changes, such as access to new treatments. For a group of patients with brain hemorrhage, who have a worse and more unpredictable prognosis, we will complement registry data with CT images.
Our interdisciplinary team, consisting of biostatisticians, AI experts, and stroke specialists, will develop, evaluate, and externally validate these innovative AI models. The knowledge generated will be crucial for identifying intervention targets at the individual, group, and healthcare management levels, facilitating resource allocation for demanding diseases like stroke. The methods developed can be adapted for the benefit of other complex disease groups.
