From AI to Machine Learning in Mental Health: In Conversation with Dr Samantha Scholtz

Image uploaded from iOS.jpg

Author: Ruchi Maniar Edited by: Isabel Wassing

Key UK government initiatives such as ‘No Health without Mental Health’ and ‘Digital First’ highlight the growing understanding that mental health and physical health go hand in hand, as well as the positive role that digital technology can play on improving mental health services. Following on from this, there has been a rapid growth in the range of psychological interventions in mental health, available online or via smartphone applications. These ‘e-therapies’ vary from wearable devices to internet-delivered therapies to apps that offer mindfulness.

We interviewed Dr Samantha Scholtz, a Consultant Psychiatrist with a special interest in bariatric surgery and weight management who believes there is a future in the use of apps supporting mental wellbeing. She is also the R&D manager for the West London Mental Health Trust, where she is responsible for developing and implementing its research strategy including expanding clinical research into dementia, forensic psychiatry, personality disorder, psychosis and the interface between physical and mental health.  Dr Scholtz started working in eating disorders and went on to become a Clinical Research Fellow, splitting her work between obesity and bariatric surgery management and research. She completed a PhD at Imperial College London, using fMRI to examine brain activity in patients that lost the desire to eat certain types of food post-bariatric surgery, and linking the response of gut hormones to these changes within the brain. This experience further cemented her conviction that physiological intervention is necessary to change perceptions and concepts within the brain. The data collected from her research were then fed back into the clinics to tailor her clinical consultations.

1.     What innovations have you seen come about in your field of work that have significantly contributed to enhancing the care you deliver?

Essentially fMRI and other tools that help support the computational analysis of data have rapidly evolved as other mathematical methods looking at behavioural and brain data are getting better day-by-day. However, I think the area where there has been the most scope for innovation is in the daily routine consultations with patients, specifically in the development of tools for physical monitoring of patients (for example mood monitoring in patients with obesity) and how these data are fed back to the patient to help facilitate changes in behaviour. The problem with applying research into clinical applications is that the process can take a very long time, sometimes up to 4 ½ years, and by the time you come up with something, it has already been replaced by something superior and the field has changed vastly.

2.     How do you see Artificial Intelligence and/or Virtual Reality contributing to your consultations that can help patients with weight management and even eating disorders?

For me what’s been interesting is learning that the more basic and concise the technology is, the easier the implementation and subsequent rate of uptake with patients. I think the role for VR could be around education and can potentially be used to bridge the clinical gaps that we may have in providing therapy. This may include setting up VR support groups or networks, as sometimes patients find it difficult to attend or don’t feel confident to attend [sessions], whereas with VR you can change your persona or look, and potentially feel more confident to share with others. AI is definitely exciting and one could even envision a world where AI allows us to provide therapy. In my opinion though, the main role of these innovations lies with their integration with physiological interventions for longer-term support.

3.     What do you think about therapeutic technology – using apps such as Pacifica and Spire to manage long-term care of health diagnoses such as anxiety, or wearable devices to monitor mood and emotions?

Long-term management would be without a doubt a good objective for which to implement such technologies, and of course a cheaper way of providing support. In my own experience with my children, who have found a way around remembering key dates using support from Siri, it is clear that young people will especially warm up to the use of interactive intelligent technology in a very human way. In general we assume that people prefer interacting with a human being, but interacting with technology can be like interacting with yourself in reality. Similar to learning how to have a dialogue with yourself really and this opens up an interesting avenue when we realise that much of what therapists do is encourage a healthy self-dialogue. Probably this will always need to be supplemented with someone providing occasional feedback in a clinical setting.

4.     Do you think there is a role for AI and machine learning in psychiatry to improve outcome or to monitor patients' progress?

I personally think machine learning is a fascinating area of innovation, and essentially what is missing in my field is the ability of being able to phenotype people – tasked with changing behaviour, medicine often takes a ‘one size fit’s all’ approach, which isn’t very ideal. I think there is a real drive to personalise treatment, considering the wide interplay of genetic, environmental and physical factors that we don’t fully understand yet, and in my view this is where machine learning can play a big role to help us predict or understand behaviour and ultimately help facilitate change.

5.     How have telehealth and telepsychiatry evolved over the last 20 years?

I think it has been slow to progress in the NHS, compared to the private sector where it’s been substantially growing. In our trust, in a service called IAPT (Improved Access to Psychological Therapy), there are pilot studies underway for clinicians to carry out Skype consultations with patients. Historically, psychologists delivered therapy that would last from 6 weeks to 3 years for all kinds of mental health disorders, and there could be substantial delays to access to treatment. IAPT has substantially eased the workload and is effectively used to manage short-term interventions. However more work is needed to use technology innovatively to increase capacity for not just therapy, but also assessment and diagnosis, crisis intervention and remote support. Ultimately, I think your generation will be the one to change this landscape dramatically by responding to the need for improved research quality and innovation.