Abstract B20: Portinari: Communicating personalized risk in cervical cancer screening using data exploration

Abstract

Background: Cervical cancer incidence rate has significantly decreased in countries that established organized screening programs. The program invites women in the age group 25 to 69 years for screening exams based on a set of guidelines. The guidelines aim to reduce over-screening of individuals at a very low-risk while effectively detecting and treating individuals at a high-risk of developing cervical cancer. However, risk determined by the screening program for a woman is often different from perception of their own risk. This contributes to a wide range of screening behavior as seen in the data collected by the Cancer Registry of Norway from 1992 to 2014. Some women get screened very frequently while some others consider themselves to be at a low-risk. Women make their own choices and we can see many patterns in their screening trajectories. Furthermore, implementation of new biomarkers may improve efficiency of the screening and requires adjustment of the guidelines. Therefore, we ask, can we use the complete set of cervical cancer screening related data collected from 1.8 million women in Norway to communicate personalized risk and concurrently evaluate performance of existing screening guidelines? Objective: Development and demonstration of a data exploration tool Portinari to communicate personalized risk of a patient based on historical data of a population. Methods and Results: We developed Portinari, a web-based, user friendly, data exploration tool for (non-)experts to query and visualize personalized risk of patients who have undergone a specific sequence of exams S. The sequence of exams of a patient is specified using a user-friendly visual editor. This visual representation is automatically transformed to a graph query. The query is executed on a graph database of screening data, which is an intuitive data structure to store and query trajectories of exams and their respective diagnosis taken over 22 years from the entire Norwegian female population. Matches for the graph query of a specific individual9s exams is a collection of identical sequences found in other patients in the database which forms the basis for risk visualization. The patient9s personal prognosis is presented by summarizing the future of all matching patients found in the database. The summary is presented as a Sankey diagram that shows arrows, representing patients flowing from one diagnosis to another with the origin being the last exam in S. The width of the arrows is proportional to the size of the represented flow which is the number of patients in our case. The Sankey diagram allows a patient to visualize both frequently taken paths taken by patients and also outliers to help them make an informed choice. We will demonstrate the use of Portinari to a) evaluate screening guidelines b) communicate personalized risk using various example scenarios. Citation Format: Sagar Sen, Manoel Horta Ribeiro, Mari Nygard. Portinari: Communicating personalized risk in cervical cancer screening using data exploration. [abstract]. In: Proceedings of the AACR Special Conference: Improving Cancer Risk Prediction for Prevention and Early Detection; Nov 16-19, 2016; Orlando, FL. Philadelphia (PA): AACR; Cancer Epidemiol Biomarkers Prev 2017;26(5 Suppl):Abstract nr B20.