Abstract
Background/Context: The online narrative of the cancer care continuum shows an alarming rise in pseudoscience-driven material; often acting to confuse patients who are in the cancer journey, especially those with inadequate information, for making various decisions on cancer treatment and care. This worrying phenomenon is co-opting more patients away from scientifically tested care strategies. Sentiment analysis refers to the use of natural language processing, text analysis, computational linguistics, and biometrics to systematically identify, extract, quantify, and study affective states and subjective information in populations. Working together, MIMOS Berhad, Malaysia's largest Applied Research and Development Centre in Information and Communications Technology, Industrial Electronics Technology and Nano-Semiconductor Technology; and the National Cancer Society of Malaysia aimed to tackle this serious health systems gap in provision of health information to cancer patients, families and the larger public. Aim: The aim of this project was to construct a sentiment analysis tool on cancer and develop an expert-driven support network for utilizing this tool for identifying pseudoscience-related cancer discussions and material in the Malaysian online space; and act to counterbalance these discussions with scientific facts. Strategy/Tactics: First, the tool was evaluated to determine its accuracy in identifying cancer-related pseudoscience news in the Malaysian context. This was done by having content experts objectively ascertain the analysis carried out by the tool. Once consensus had been reached with the experts, the tool was deployed. The support system mechanism consisted of a voluntarily recruited expert panel consisting of healthcare professionals briefed on the mechanism of the tool. Once the tool was deployed, the resultant analysis was shared out to the expert panel who then responded; counterbalancing the pseudoscience material identified in the respective online medium with accurate information. Program/Policy process: The sentiment analysis tool deployed machine learning, statistics, and natural language processing techniques to automate sentiment analysis on large collections of online cancer-related texts, including Web pages, online news, Internet discussion groups, online reviews, Web blogs, and social media. The tool was combined with a support-systems network of healthcare professionals who acted on the analysis results. A weekly run analysis and feedback mechanism was determined to be viable in terms of turn-around-time (TOT) while still remaining 'current' in the fast-paced online scene. Outcomes/What was learned: A marriage of an artificial intelligence system backed by human content experts can be a viable, sustainable mechanism in reducing the impact of pseudoscience in the online cancer ecosphere and help in provisioning accurate health information to cancer patients, families and the general public as a whole.
Published Version
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a call