EDISCIENCES

ABSTRACT

Background Online cancer communities provide a space for cancer patients and their caregivers to discuss issues related to the disease itself. On their institutional website, the Italian Association of Cancer Patients offers visitors a forum pointing it out as a free space where it is possible to share one’s story, exchanging emotions, feelings, information, and thoughts through writing. This study aims to explore, within the AIMaC community and regarding prostate cancer, how member behavior is distributed and how message sentiment changes over time. Materials and methods We retrospectively analyzed all Prostate cancer-related discussion posts recorded within the AIMaC community forum from 2010 to 2019. We apply to the dataset a sentiment analysis using an unsupervised approach of Natural Language Processing Results Three thousand ve hundred forty-nine messages exchanged by 219 members belonging to the prostate cancer posts were extracted from 2010 to 2019 for 39,040 words. The authors explored the time series trend for each sentiment. By dividing the original dataset into two sub-datasets organized by the rst and second ve-year periods, this insight allowed us to apply the Student’s t-test to the differences between these two new time series for both negative and positive feelings. The results is t(9)=-7,664228406, p<0,0001. Conclusion The proposed model may allow even more accurate text mining analyses to evaluate data such as unmet needs and any other patient-related concerns. These data could be helpful to healthcare professionals to consolidate even more the management of the physician-patient-disease system.