Purpose and objectives of the study
Omics-driven Biomarker Discovery
Omics has attracted much attention to support discovery of novel biomarkers. However, of the estimated 1,000,000 biomarkers identified in omics studies only a small fraction (<1%) has achieved FDA approval. Failures in marker development equate to lost resources, from consuming money, time, labor, talent, and credibility for the field. A major bottleneck in development of omics-informed biomarkers is the discovery phase, which often lead to biomarkers with performances that fail to meet the demands of the clinic. The combination of biomarkers and machine learning can led to enhancement in accuracy-performances with at least 10%, reducing the risk of failure in clinical validations. We aim to use Artificial Intelligence based strategies to efficiently mine big biomedical data and quickly assess combinations of biomarkers from large-scale omics data. In particular, we apply advanced machine learning and heuristic techniques to address two major unsolved problems in biomarker discovery: (i) the selection of the smallest combinations of molecular signatures that still retains high predictive performances and (ii) the integration of biological knowledge, clinical and multi-omics data in a systems biology framework to significantly increase the accuracy and robustness of biomarker models for chemical clinical applications.
Computer-aided drug target discovery
Selecting the right biological target or a combination of targets is the first fundamental task for any successful drug development process. However, the advent high throughput technologies enabling comprehensive analysis of genes, transcripts, proteins and other relevant biological molecules has dramatically increased the number of promising drug targets. We aim to apply network-based approaches to systematically compile efficacy and safety scores of putative drug targets from single- and multi-omics data. These scores are used select drug targets representing the best trade-offs between efficacy and safety.
The Significance of Research
Our research projects will provide advanced AI-based technologies to make personalized medicine a more tangible reality. Novel AI-based solutions will translate the huge amount of biomedical data available in public data repositories into effective clinically actionable tools to aid medical decision-making. These tools will be part of larger system transformation in which emerging technologies – genomic, digital, robotic, AI – will come together to provide a greater capacity for prediction and prevention in a clinical setting.
European Commission, University of Eastern Finland
Markku Laakso (UEF), Mikko Hiltunen (UEF), Jorma Palvimo (UEF), Anna-Liisa Levonen (UEF), Jussi Paananen (UEF), Minna Kaikkonen-Määttä (UEF), Alina Solomon (UEF), Einari Niskanen (UEF), Dario Greco (University of Tampere), Harri Alenius (Karolinska), Nanna Fyhrquist (Karolinska), Piia Karisola (University of Helsinki), Francesco Napolitano (TIGEM), Ferdinando Bonfiglio (University of Basel).