Publications » Risk Prediction as a Service: a DSS architecture promoting interoperability and collaboration

Risk Prediction as a Service: a DSS architecture promoting interoperability and collaboration

Stefano Mariani, Franco Zambonelli, Akos Tenyi, Isaac Cano, Josep Roca
Clinical research and practice are rapidly changing mostly due to Information and Communication Technology, especially, as Machine Learning (ML) offers great potential for predictive and personalised medicine. Nevertheless, barriers are still existing for widespread adoption of ML tools, as highlighted by studies from the European Union. In this paper, we propose an architecture for a Decision Support Systems assisting clinicians in assessing health risk of patients by delivering “Risk Prediction as a Service”. By leveraging standard web technologies as well as the PMML and PFA formats for exchange of trained models, we achieve ubiquitous access to predictions, ease of deployment, seamless interoperability, while promoting collaboration.
Keywords: risk prediction; PMML; PFA; machine learning; decision support system; CONNECARE
CBMS 2019, 2019
@article{,
	year = 2019,
	keywords = {risk prediction; PMML; PFA; machine learning; decision support system; CONNECARE},
	status = {Submitted},
	venue_list = {--},
	venue_s = {32th IEEE International Symposium on Computer-Based Medical Systems},
	author = {Mariani, Stefano and Zambonelli, Franco and Tenyi, Akos and Cano, Isaac and Roca, Josep},
	title = {Risk Prediction as a Service: a DSS architecture promoting interoperability and collaboration},
	abstract = {Clinical research and practice are rapidly changing mostly due to Information and Communication Technology, especially, as Machine Learning (ML) offers great potential for predictive and personalised medicine. Nevertheless, barriers are still existing for widespread adoption of ML tools, as highlighted by studies from the European Union. In this paper, we propose an architecture for a Decision Support Systems assisting clinicians in assessing health risk of patients by delivering “Risk Prediction as a Service”. By leveraging standard web technologies as well as the PMML and PFA formats for exchange of trained models, we achieve ubiquitous access to predictions, ease of deployment, seamless interoperability, while promoting collaboration.
}}