Social Media Analytics Leveraging Deep Learning Models for the Study of Opioid Addiction: Perception, Pattern and Acceptance

In this project, we aim at three specific objectives of using social media analytics leveraging deep learning models to facilitate Medication Assisted Treatment: (1) Public perception and stigma about MAT for opioid use disorder patients. For layman, the difference between drug and medication is subtle. We propose to use social media platform such as twitter to study public perception and stigma of methadone and buprenorphine in MAT. (2) Non-prescription methadone and buprenorphine use pattern discovery. Overdose on methadone is often lethal and there have been multiple reports of accidental death from inappropriate use of methadone with or without other drugs. In this project, we would like to use social media to find out illicit use pattern of these two FDA approved maintenance medications – e.g., What other drugs do they often concurrently use? Do they use buprenorphine or methadone to "get high," or to self-medicate? Do people use it as a bridge treatment until they are able to access further help? Is the experience positive or helpful in seeking further treatment? (3) Treatment setting, patient acceptance and outcome analysis. To the best of our knowledge, little is known about the relationship between treatment setting and patient acceptance, and how this might affect a patient's decision to seek recovery through MAT. We propose to devise novel deep learning models based on social media data to help us better understand fundamental issues related to treatment decision.