Professor Andrew McIntosh describes what Mental Health Data Science is and introduces the four research approaches being undertaken by his team at The University of Edinburgh:
1) Enhance Existing Studies
2) Mining Information
3) Genetics and Genomics
4) Public Engagement
Mental Health Data Science Scotland is a collaboration between the universities in Scotland, the NHS and between other providers of relevant mental health information; so that might be information on education, on poverty and other areas.
Mental health data science is the intersection of four specialisms. The first three are mathematics statistics and computer science. The fourth aspect of mental health data science is in the treatment of mental health, and that could include people from psychiatry, clinical psychology, psychology, humanities or the arts.
We use four general approaches in our collaboration.
The first of these is to enhance existing research studies, the second of these is mining information from the medical and other records, the third is the use of genomics and the fourth is public engagement.
Firstly we're looking to enhance existing studies, because, there are some very large studies out there, that have not so far shown a great deal of interest in mental health research, but to which we can add (at low cost) information about people's mental health, by sending them a questionnaire by a post or by sending to their email a questionnaire that they can fill out online
The second approach that we're taking in our collaboration is mining information that has been collected for other purposes. The simplest example of this is when you go to your GP with a mental health problem, perhaps anxiety or stress. That information is recorded for the purposes of clinical care. Under some circumstances, however, that information is recorded and can be very useful for mental health research.
There's also other information out there that's of value for mental health research: information on your birth, on your employment, information that's relevant to poverty that's also out there and relevant to mental health. And we seek to also capture some of that information by something called record linkage, where we were able to retrieve people's medical records and analyze them,
And also once we have that information, sometimes it's extremely large and we develop computer algorithms to help us mine that information in a non-subjective and bias-free manner.
One of the approaches that we're using in our collaboration is the better and smarter use of genetics in our research. So genetic factors contribute to all mental health disorders to a smaller or to a major degree. The thing that's really exciting about using genetic information is the fact that it is fixed from birth. So if you find an association between a change in your genetic makeup and mental health, it automatically gives you some information on the cause. So we think that's really important because it enables us to have an anchor against which we can test whether factors that are associated with mental health disorders are a cause or a consequence of the illness.
A second way in which we're using genetic information is to develop new treatments. So we think that the genetic information that's associated with mental disorder is especially important in determining what new and more effective drugs might be available in the future for the treatment of your condition.
A major strand of our activity around mental health data sciences is in Public Engagement. Much of the data that we seek to obtain in our research is very sensitive. It could be around the treatment you've received by your GP for mental health difficulties, or your employment status, or where and how successful your education was, earlier on in your life. Now that information we believe is very valuable for mental health research, but it's obviously very sensitive information. We take very seriously the degree to which we have to keep it private and confidential. So what we're aiming to do in our collaboration is to take the public along with us; to use them to guide our research, to help highlight important questions for them and to make sure that our enthusiasm for analysing the data doesn't outpace their permission and willingness to share it.