Annual Meeting Abstracts: View

<< Back to Search Results

(Abst. 1.283), 2019

Artificial Intelligence (AI) to Understand Suicidality Among Teenagers and Adults Suffering from Epilepsy: The Value of Digital Conversation to Understand Their Mindset
Authors: Tatiana Falcone, Cleveland Clinic; Anjali Dagar, Cleveland Clinic; Ruby Castilla- Puentes, Johnson and Johnson; Amit Anand, Cleveland Clinic; Caroline Brethenoux, Cultureintel; Liliana Gil Valleta, Cultureintel; Elia Pestane-Knight, Cleveland Clinic; Jane Timmons-Mitchell, CWRU
Content: Rationale: Although around 30-50% people with epilepsy (PWE) suffer from depression (1), which is a risk factor for suicidal behavior, not much is known about the major motivations for suicidal thoughts among teenagers and adults suffering epilepsy. The aim of our study is to fill this gap through the analysis of online conversations. Methods: CulturIntel®, uses big data and AI suite of tools to mine digital conversation/comments including suicide, suicidal thoughts, intent, plan, attempt etc. Using ‘scrape and listen’ tools, open-source online conversations from self-identified teenagers and adults users online who endorsed suffering from epilepsy were analyzed. The search was limited to conversations from the USA internet protocol (IP) addresses in a 12-month period (September 26, 2017 through September 25, 2018). The unstructured big data was examined using natural language processing and text analytics for previously described and undescribed patterns in the data under human-assisted repeated training, testing, and reviewing of the program output. Qualitative statistical analysis was performed to analyze the final available data. Results: A total of 222, 000 conversations were collected and analyzed. This included 181,000 conversations by adults and 41,000 by teenagers of which 5800 (3%) and 3200 (8%) posts, respectively, were related to suicidal thoughts. The content analysis showed that 30% of posts by teenagers discussed social consequences of seizures compared to 21% posts by adults. Another 29 % of posts by teenagers were looking for emotional support to deal with the impact of the illness compared to 19% adults. In contrast, concerns about physical impairments from seizure were expressed in 29% posts by adults compared to 21% teenagers. Almost two-thirds (63%) of teenagers expressed fear about the ‘unknown’ compared to only 12% adults. Ten-times more adults expressed attitude of having ‘given-up’ compared to teenagers. Analysis of the primary sentiment behind the conversations was negative in two-thirds of adults and teenagers. Conclusions: Our analysis of digital conversations among PWE, through machine learning technology, shows that a small but significant percentage of teenagers and adults express suicidality online. Such behavior is more common among teenagers. This first of its kind research highlights the differences in concerns and the shared sentiments among teenagers and adults suffering suicidality related to epilepsy. Funding: Project IMPACTT from HRSA
Figure 1
Figure 2