Super Excited to have two journal acceptances on:
- Knowledge-infused Abstractive Summarization (JMIR)
- Characterization of Time-variant and Time-invariant Assessment of Suicidality on Reddit using C-SSRS
The study also shows medical knowledge utilization in contextualizing and abstracting the noisy discourse on social media and clinical diagnostic interviews.
The first study on summarizing open-ended and template-agnostic clinical diagnostic interviews enumerated a sequence of mental health challenges and how explainable AI can help develop plausible solutions. It highlights an approach to effectively summarize 57 sentences long interviews to 7 sentence summaries contains “Essential Questions asked by Clinicians” and “Meaningful Responses provided by patients.”
The second study on understanding suicide risk assessment on Reddit touches upon two fundamental problems when dealing with social media:
- It is not sufficient to estimate suicide risk only through longitudinal machine learning modeling.
- A user with a history of indulging in suicidal behavior, upon receiving timely care, becomes a supportive/helpful person on Social Media. Identifying varying user-types and content-types while maintaining the outcome’s clinical relevance is the urgent challenge addressed in the research.