Keynote

Titel: KI-basierten und datengetriebenen Analyse von Pandemiedaten

Vortragender: Dr. Sobhan Moazemi, Fraunhofer-Institute for Algorithms and Scientific Computing (SCAI)

Abstract: The recent COVID-19 pandemic has highlighted the urgent need for innovative solutions to combat such global health emergencies. Artificial intelligence (AI) has emerged as a powerful tool in this fight, offering unique capabilities for analyzing vast amounts of data, identifying patterns, and making predictions that can inform critical decisions.

Our research focused on utilizing these capabilities of AI in the fight against COVID-19. We developed an AI model predicting patient mortality with high accuracy and identified dementia-related pathways involved in severe cases. We showed that molecular mechanisms related to dementia, one of the relevant predictors in our model, intersect with those associated to COVID-19. Most notably, among these molecules was tyrosine kinase 2 (TYK2), a protein that has been patented as drug target in Alzheimer’s Disease but also genetically associated with severe COVID-19 outcomes. This led us to explore existing drugs targeting these pathways, discovering potential treatments with existing anti-cancer drugs.

Furthermore, the World Health Organization (WHO) emphasized the critical need for new medications alongside vaccines to combat the COVID-19 pandemic. One promising strategy involves identifying target proteins that could be effectively disrupted by existing drugs, potentially offering rapid therapeutic options for COVID-19 patients. One possible strategy is to identify target proteins, for which a perturbation by an existing compound is likely to benefit COVID-19 patients. To this end, we created a web tool, namely, GuiltyTargets-COVID-19, to identify novel drug targets for COVID-19. This tool analyzes gene expression and protein interaction data to prioritize potential drug targets and assess their druggability, considering potential side effects of existing drugs targeting them.

Lastly, machine learning (ML) models hold promise for easing the burden on healthcare systems during pandemics such as COVID-19. They can analyze routinely collected electronic health records (EHRs) to identify patients at high risk of severe disease progression. However, EHR data presents challenges due to irregular timestamps. As another ML-based solution, we introduced ExMedBERT, utilizing transformer-based models, extending the Med-BERT model to incorporate additional patient information and improve risk prediction for COVID-19 patients. This model outperformed other models and provided insights into the risk factors.

In conclusion, these studies showcase the diverse applications of AI and ML in the fight against COVID-19, ranging from patient risk prediction and drug discovery to repurposing existing medications.

Short Bio: Dr. rer. nat. Sobhan Moazemi is Senior Biomedical Data Scienctist at the Fraunhofer-Institute for Algorithms and Scientific Computing (SCAI) AI & Data Science Group, Department of Bioinformatics in Sankt Augustin. With a background in data science and medical AI, Dr. Sobhan Moazemi joined Fraunhofer SCAI as a Senior Postdoc Biomedical Data Scientist in October 2023. His research mainly contributed to the applications of AI and ML in personalized medicine. In his current role, he is responsible for project management and conducting applied research in generative AI and synthetic health data.