Online peer-to-peer support platforms such as TalkLife enable conversations between millions of people who seek and provide mental health support. If successful, web-based mental health conversations could improve access to treatment and reduce the global disease burden. Psychologists have repeatedly demonstrated that empathy, the ability to understand and feel the emotions and experiences of others, is a key component leading to positive outcomes in supportive conversations. Computational methods that can empower non-expert peer-supporters with empathy-based feedback and training has the potential to help them express higher levels of empathy and in turn, improve the effectiveness of online support platforms.

In this talk, I will present our work on understanding and improving empathy in online mental health support conversations. First, we design a novel framework and dataset of empathic conversations and develop a RoBERTa-based bi-encoder model for identifying empathy in conversations. Using this model, we demonstrate that highly empathic conversations are rare on online mental health platforms. Therefore, to improve empathy in conversations, we introduce Empathic Rewriting, a new task that aims to transform low-empathy conversational posts to higher empathy. We propose PARTNER, a deep reinforcement learning agent that learns to make sentence-level edits to conversations in order to increase the expressed level of empathy while maintaining conversation quality through fluency, specificity, and diversity.

Case Study Summary:

  • The scientific problem we tackled:
    Access to mental health care is a global challenge that online peer support platforms like TalkLife can help mitigate. Millions of people seek and provide support online but they struggle to have effective conversations. We developed computational methods to help facilitate empathic interactions in peer support platforms through intelligent and actionable feedback. This research helped us understand the role of empathy in peer support and how to empower peer supporters in writing more empathic responses.
  • The computational methods we used:
    We trained Natural Language Processing and Reinforcement Learning models for identifying and improving empathy in peer support conversations. Using these models, we have designed interactive tools for providing intelligent and actionable real-time feedback to peer supporters and for suggesting concrete ways in which the conversations can be made more empathic.
  • The cloud resources we used:
    We used Azure GPUs for training our models and deploying them in the real world.
  • The differences we’ve observed between locally-provided and cloud-provided resources:
    Cloud-provided resources provide significantly more flexibility and scale than locally-provided resources. Training large AI models requires GPUs with varying memory needs. Azure facilitates instantiating different types of GPUs. Also, for the models to optimally learn these complex tasks, we needed to perform a hyperparameter tuning using grid search. This requires training the same model several times. While this is infeasible on a local machine, we can scale this hyperparameter search by creating multiple Azure GPUs.

Author Bio:

Ashish Sharma is a PhD student at the Paul G. Allen School of Computer Science & Engineering, University of Washington. His research focuses on building computational techniques for encouraging empathic conversations in text-based mental health support. Ashish holds a dual degree (B.Tech. + M.Tech.) in computer science from the Indian Institute of Technology, Kharagpur and has worked as a Research Fellow at Microsoft Research, India.

For further information:

  • http://bdata.uw.edu/empathy/

RRoCCET21 is a conference that was held virtually by CloudBank from August 10th through 12th, 2021. Its intention is to inspire you to consider utilizing the cloud in your research, by way of sharing the success stories of others. We hope the proceedings, of which this case study is a part, give you an idea of what is possible and act as a “recipe book” for mapping powerful computational resources onto your own field of inquiry.