AI can also be used in innovative ways to reveal mental health trends in datasets from digital mental health support services, which can be of great value not just to those services, but also to other organisations, researchers, government, and the general public. Moreover these approaches offer powerful ways to support service evaluation and innovation.
At Mental Health Innovations, we are using natural language processing and machine learning, in partnership with the IGHI and others, to capitalise on these opportunities. Using machine learning models, we are gaining a better understanding of factors driving patterns of demand for the service, from social media posting to weather dynamics; we have explored potential demographic biases in our data, ensuring that we gain a more thorough understanding of how different groups of people are using the service; and we have conducted large scale analyses of service user feedback to understand topics and themes that help us to improve service provision.
Thanks to funding from Google, we are transforming the way we evaluate and understand the quality of the Shout service, by deploying machine learning models to provide comprehensive and diverse mechanisms for monitoring service quality at scale. For example, one key aim of any conversation at Shout is for the texter to feel more calm. How can we assess this at scale on a daily basis? We have started to use machine learning models to track relevant conversation features and also to give us insights into the characteristics of ‘successful’ conversations, so that we can feed this back into our ongoing training and practice.