An innovative new pilot study from the think tank Demos, in partnership with The King’s Fund, reveals an enormous untapped resource of public online forum discussions as a possible resource for policy makers and practitioners in healthcare.
43% of internet users have now used the web to access health information, up from 18% in 2007. Alongside well administrated official sources like NHS Choices, unregulated online forums have grown to be valuable spaces for users to discuss conditions and treatments with those who have had similar experiences, ask questions, and share advice.
In this new study, the Centre for the Analysis of Social Media at Demos developed and utilised pioneering technology to analyse over one million posts on mental health issues.
The study identified these forums as a rich source of qualitative data, ranging from discussions of medication side effects to detailed patient opinions on hospitals and treatments. These remain largely untapped by healthcare professionals, however, due to the difficulty of analysing such vast quantities of linguistic data. The paper also examines in detail how such data can be accessed and used in a way that could improve understanding of conditions and support, but remains ethical and respects user privacy
This pilot study from Demos and The King’s Fund demonstrates how such forums can provide insights to the healthcare industry through innovative technology. Posts on publicly available pages were “read” by a modified web scraper and anonymised. Researchers then trained bespoke ‘Natural Language Processing’ algorithms, a form of artificial intelligence, to answer specific questions about how users discuss mental health online.
- Cries for help
The study found a significant number of posts were users reaching out to other members for urgent assistance with their mental health issues. A key advantage of forums is that they are available in a person’s home when other forms of assistance might be unavailable. Many of these users mentioned having made contact with health professionals, and analysing these further could provide huge potential to flag up failing services and common complaints.
- Cognitive behavioural therapy
The piloted software had a high accuracy rate (72%) in identifying posts describing experiences of patients who have a personal experience of CBT, a form of therapy. The process was able to identify a wide range of narratives involving the effectiveness, perception of and demand for this treatment. By searching for specific keywords in this way, there is strong potential for policymakers to identify public conversations about particular treatments or medications.
Many people experience mental health problems alongside long-term physical conditions. This software approach was able to identify posts discussing mental health issues in conjunction with other issues (respiratory disease, diabetes and arthritis), providing insight into the experience of patients with physical and mental health problems interacting with the health system.
The successes of this pilot demonstrate that such unstructured data sets are amenable to analysis with the right technology, which would be advantageous for:
- Forum owners: enabling them to better understand the issues discussed on their sites and the possibility to offer better targeted information or relationships with service providers.
- Service providers: helping them develop a deeper understanding of users’ experiences, allowing for a more thoughtful design in response.
- Health regulators: giving them additional insight into the performance of different providers.
Commenting on the findings, Josh Smith, Researcher in the Centre for the Analysis of Social Media at Demos, said:
“This pilot project demonstrates that public online health forums could prove to be a valuable resource for policymakers and practitioners who want to better understand the experiences of often hard to reach communities. It highlights the potential for new technology and methodologies to provide a whole new perspective on mental health, while laying the foundations for this work to be conducted in an ethical and effective manner.”
The full paper and methodological notes are available here.