Mental health therapy analysis for better patient outcomes

Toni Sekinah, Research analyst and features editor, DataIQ

With 10th October designated as World Mental Health Day, a lot of attention has been drawn to the importance of mental wellbeing and ways in which sufferers of mental illness can be supported and treated. Ieso Digital Health has a commitment to using science to drive change in mental healthcare and, since it was founded in 2000, has been offering one-to-one cognitive behaviour therapy to patients with common health conditions such as depression and anxiety.

Sad woman at laptopSessions with BABCP-certified therapists are delivered online through a computer, smartphone or tablet, which is convenient for the patient and also allows the therapist to monitor progress.In addition, the digital delivery of therapy provides the in-house team of researchers with a lot of data to look at.

Eleven researchers from Ieso, the University of Cambridge, King’s College London and the Vanderbilt University have just published an article in the British Journal of Psychiatry looking at the factors that correlate to successful treatment outcomes.

"We had a hunch and we wanted to confirm this in a real-world data set."

Ana Catarino, senior scientist at Ieso Digital HealthLead author and senior scientist, Dr Ana Catarino, who managed the study said that they have a large data set and they are committed to learning from it. “We really wanted to use that data to get insights on why patients respond differently to treatment. Not all patients respond equally and we wanted to know what was driving those differences.”

The researchers had a hunch that the longer a patient spends waiting for assessment, the lower their likelihood of recovery. “We wanted to confirm this in a real-world data set.”

This was demonstrated through positive responses to treatment being associated with shorter waiting times for initial assessment as well as longer treatment durations in terms of the number of sessions. Also, patients with long-term physical conditions as well as those with high severity scores at assessment are less likely show good clinical outcomes.

Furthermore, they found that there was a positive correlation between age and a positive clinical outcome, in contradiction to their expectation. Dr Valentin Tablan, one of Catarino’s co-authors and senior vice president for artificial intelligence at Ieso, said: “We assumed that the younger patients would be more likely to engage with this kind of modality and that it would be more difficult for the older patients. What we found was the opposite. You shouldn’t just make assumptions.” However, to put this result into context, Catarino said that this perhaps indicated a difference in severity of symptoms across age groups.

"For privacy and security, we didn't work with data outside of our systems."

Dr Valentin Tablan, SVP of AI at Ieso Digital HealthTablan explained that all the “unglamourous” work of collecting and cleaning the data, as well as the analysis which was done in R, took in place in-house at Ieso. This was to ensure the anonymity of the patient data and make sure that their privacy was respected.

“We pay great attention to data from a security and privacy standpoint. That is one of the reasons we didn’t work with data outside of our systems,” he said. The growing in-house team comprises scientists, practising clinicians, data analysts and software engineers.

Catarino stated it was challenging to work with such a volume of data and look at as many service and patient variables as possible without compromising the integrity of the analysis. The data analysed came from patients who received internet-enabled CBT for the treatment of a mental health disorder between April 2015 and March 2016 with a sample size of 2,211 patients. The analysis took about three months.

For the lead author, this study represents an important first step in showing the value of data and data-driven insight into learning what works in psychotherapy and using that information to drive change in services.

She elaborated: “By driving change, I mean more personalised treatment protocols that can be delivered to patients, so that each patient gets the right care for their condition and not a one-size-fits-all treatment that is currently offered to patients with mental health problems. So, we hope that this is a milestone study towards this goal.”

"We're working towards a world without mental illness."

Robot finger on laptop keyboardThe researchers are currently working on a study that will look at how what happens in the virtual therapy room correlates with outcomes for the patients. Tablan stated: “We are starting to use artificial intelligence to analyse the contents of therapy sessions and identify the active ingredients used by the therapists and find out what works best with which patient cohort.”

“Our stated goal is to defeat mental illness and that’s what we’re working towards - a world without mental illness and there’s all kinds of things one can do can help in this direction,” said Tablan. Data science is one of those things.