Our Research

DECEMBER 2022

Improving uptake of mental health crisis resources (Preprint)

How might we increase the effectiveness of online crisis referrals? We compared our referral system with a standard link to 988/Lifelines and found significant reductions in hopelessness. People who received our referral were also more than twice as likely to report using the resources provided.

JUNE 2022

Embedding single-session interventions into online social networks

We embedded single-session interventions directly into online social networks, allowing us to reach users where they are. We found significant pre-post improvements in hopelessness, self-hatred, and motivation to stop self-harming. (In press, JMIR - Formative)

JANUARY 2019

Machine learning-based crisis triage

We developed methods to increase utilization of crisis resources such as The National Suicide Prevention Lifeline. In a randomized controlled trial with Harvard, we showed our methods increased conversion rates by 23%.

OCTOBER 2018

Linguistic correlates of effective online peer support

Koko has millions of posts and responses from its peer support network. We analyzed linguistic patterns that predicted favorable outcomes.

JUNE 2018

Can we build an AI therapist? Can we build empathy from algorithms?

We created an artificially empathic chatbot using Koko data. We find that, while we can simulate empathic expression to a degree, there may be an upper bound machines can never cross. Unlike humans, they don't have lived experiences and so their efforts to empathize might always ring hollow.

MAY 2017

Helping others on Koko predicts decreased symptoms of depression

Most people looking to get help on Koko go on to help multiple other people. We make it simple and easy. We also find that helping others on Koko has a significant effect on mood regulation and depression symptoms.

MARCH 2015

Randomized controlled trial of the Koko platform

We randomly assigned people to use Koko (then called 'Panoply') or a matched control task (expressive writing). We found improvements for both, but the differences were significantly greater for the Koko platform across all outcome measures.