Cancer accounted for nearly one in every four deaths in the United States and approximately 13% or 7.4 million deaths worldwide in 2007 [1]. Many survivors and their family or friend caregivers experience not only physical effects but also emotional effects such as stress, anxiety, and depression from cancer and cancer treatment. Due to the worldwide explosive usage of internet for social interactions (e.g., through social media and online social networks), a rapidly increasing number of cancer survivors seek social support by participating in an online health community. What is the impact of these online interactions to the quality of life of cancer survivors? How can these interactions be further leveraged to better address needs of cancer survivors? Motivated to address questions such as these, an interdisciplinary group of faculty from the Pennsylvania State University and Purdue University launched the Cancer Informatics Initiative in 2010. The initiative is affiliated with the Penn State Hershey Cancer Institute under Cancer Control Research Program and collaborates with Community Sciences and Health Outcomes Core, which has developed the Northern Appalachian Cancer Network.

The mission of the initiative is to: (1) improve our understanding about the information needs of cancer survivors (which includes cancer patients, their family, and friends), and how these needs are supported by information flows in online community such as the American Cancer Society’s Cancer Survivors Network (CSN); (2) develop models about the influence and roles of participants, spreading of opinions and rumors, formation and resolution of conflicts, and the growth and the health of an online cancer support community; and (3) innovate, design, and deploy next-generation search engines, question answering systems, and information recommendation systems for cancer-survivors online communities that deliver high quality and timely information that addresses the information needs of cancer survivors.

Through partnership with the American Cancer Society, the Initiative has studied the impacts of online interactions to members of Cancer Survivors Network. The project developed a novel approach using machine learning techniques to automatically label the sentiments of all posts in CSN’s online forum (about half a million posts collected over ten years since 2000). By analyzing the sentiment change of those who initiated threads of discussion in the forum, the study indicated that more than 90% of the thread initiators change their sentiment in the positive direction due to responses posted by others on the thread. Our study revealed, for the first time, that the number of participants and their average sentiments can affect the sentiment changes of thread initiators. More specifically, increasing the number of participants on a thread increases the likelihood that the thread initiator has a positive sentiment change. Furthermore, the more positive the average sentiment of participants on a thread is, the more positive is the sentiment change of the thread initiators. To the best of our knowledge, ours is the first systematic study about information exchanges in a large-scale health online forum. It not only reveals important new insights about the effect of online interactions to the sentiment of people, but it also lays the foundation for fine-grained sentiment analysis, which is important for detecting and tracking the opinion, the information need, and the perception of cancer survivors. Further information about our research can be found from our publications. If you would like to reach us, please send e-mail to cani@ist.psu.edu.

The research activities of the Cancer Informatics Initiative have been supported by the following sponsors:

1. American Cancer Society Collaboration Agreement on CSN Online Discussion Analysis
2. Seed Grant from Social Science Research Institute, Penn State University
3. Seed Grant from Penn State Cancer Research Institute, Population Health and Cancer Control Research Programs

[1] ACS, American Cancer Society Cancer Facts and Figure 2011, 2011.