Ontologies can be a very useful means for organizing knowledge about a domain. In particular, the formal process of designing an ontology makes it appropriate for setting standards regarding the organization of the domain knowledge. One of the reasons that ontologies have become of interest recently is the rise of the semantic web, and the fact that ontologies can be defined using the protocols and standards therein.  This provides a number of opportunities for leveraging the power and computability of the semantic web. In particular, integrating the domain knowledge with other domains without needing to duplicate effort is a key advantage of developing an ontology.

We have four key goals in this session: (a) to provide an overview of ontologies, their use, and their background, (b) to describe the relevant elements of the semantic web and the software used to generate and apply ontologies, (c) to provide further insight by presenting several ontologies that have been developed for the radiation oncology field, and (d) to describe the advantages of a directed effort to develop appropriate and useful ontologies for radiation oncology.

The material described in this session should fulfil a number of objectives:

• To be a bridge between interested physicists who have little formal background in the field to the medical physics community that has been working on ontologies;

• To educate interested parties in some detail regarding the generation and use of ontologies;

• To inform others working in the informatics field about this initiative;

• To help to integrate efforts in the radiation oncology informatics field to further the use of ontologies in radiation oncology.

Session duration: 70 minutes, 3 x 20 minutes for talks, 10 minutes for questions.

Predictive modeling using electronic health record (EHR) and medical images is anticipated to drive personalized medicine and improve healthcare. With the emergence of statistical learning, artificial intelligence oncology applications such as early detection/classification, risk stratification, automated planning, care adaptation and workflow optimization could be developed using historical and new data. However, the large majority of the health information is currently trapped due to institutional concern of data privacy and poor data infrastructure. Translational research in data science will face significant challenges to thrive for the ‘4Vs’ of ‘Big Data’; veracity, velocity, variety, and volume of data. In this session, we will discuss current challenges and possible solutions for unleashing oncology data for precision medicine. We will review the importance of patient participation for systematic collection of patient-reported outcomes. Finally, a possible framework will be presented to dynamically summarize key features of a patient profile along and after the cancer care journey. Examples of ongoing and future research efforts using clinical variables, texts and images will be discussed.