Abstract Body

Introduction

Digital mental health services have been a part of routine care at a few locations worldwide since almost 15 years, most often in the form of Internet-based Cognitive Behavioural Therapy (ICBT) with scheduled weekly therapist-guidance. Personalization in the form of individual tailoring of treatment content is promising in ICBT. Digital Self-care, interventions constructed to be self-guided, would need to be constructed carefully to achieve equal levels of adherence and symptom reductions compared to therapist-guided interventions, especially when including individually tailored content.

Objectives

To construct an individually tailored self-care intervention including a technical solution, acting as a proof of concept that self-guided digital interventions for mental health can be administered in a safe, effective, personalized and cost-effective way.

Methods

In step I, a new digital platform is created based on the experience from previous successful implementations of ICBT together with experts on user experience. A series of digital mental health tools based on ICBT are tested for safety, usability and credibility. In step II these tools are combined into individually tailored package interventions for different conditions and optimized for greater efficiency. In step III these optimized interventions would be compared to their counterpart therapist-guided interventions in randomized trials.

Results

Preliminary results from step I will be presented, including the current development of the digital platform and feasibility data from the first three studies.

Abstract Body

Guided self-help interventions are effective in treating symptoms of various mental disorders, including depressive, anxiety, and posttraumatic stress disorders. Research also suggests that these interventions may be effective for refugee populations. However, proportion of drop-out and non-response are substantial, especially in this highly vulnerable group of patients. Tailoring treatments to the individual patient may be an important step towards improving patient-treatment fit and may help to increase success rates. While tailoring can be easily realized in face-to-face treatments, it becomes more complex in Internet-based treatments where treatment sequences are usually defined in advance.

In this talk, we will present our theoretical considerations and decisions regarding the tailoring process in a randomized-controlled comparison of transdiagnostic CBT for refugee patients in an online versus face-to-face format. The trial will include N=320 Arabic speaking patients suffering from an emotional disorder. The transdiagnostic treatment includes modules for symptoms of depression, anxiety, substance abuse, post-traumatic stress, aggression, and suicidal ideation. Modules are tailored to the specific patient. We will discuss who or what should inform the tailoring decision (patient, therapist, questionnaire data, diagnostic interview) and when tailoring decisions should be made (prior and/or early and/or later in treatment). We will present options of how tailoring decisions can be standardized and be kept comparable in different treatment formats. We will present our first experiences with tailoring treatment modules to severely impaired and highly comorbid patients.

Abstract Body

In 2012, the Danish government introduced a national action plan for telemedicine comprising, amongst other initiatives, Internet-based Cognitive Behavioural Therapy (iCBT) for adult depression later also including anxiety. First introduced by Dr. Selmi in 1991 iCBT has since been researched extensively with positive results - even comparable to CBT delivered face-to-face. However, not much is known about the effectiveness once iCBT is implemented in routine care.

During this presentation, the nationwide iCBT clinic Internetpsykiatrien.dk is described as well as the results from the first cohort of patients undergoing treatment.

A naturalistic cohort design was used including patients from April 1st, 2016 to April 1st, 2017. Primary outcomes were PHQ-9 for depressed patients and GAD-7 for anxious patients measured pre- and post-treatment. Primary analyses were conducted using a linear mixed effects model with random slope and intercept. Results were benchmarked to published effectiveness and efficacy trials of guided iCBT.

A total of N=203 patients were included in the analyses (depression n=60, anxiety n=143), mainly female (depression 78.3%, anxiety 65.7%) with a mean age of M=36.03 (SD 10.97) for the depressed patients and M=36.80 (SD 13.55) for the anxious. The primary analyses revealed large and significant reductions in the symptom levels of depression (beta=-6.27, SE 0.83, P<.001, d=1.0) and anxiety (beta=-3.78, SE 0.43, P<.001, d=1.1).

These results appear to support the hypothesis, that iCBT can be clinically effective even in a routine care setting. This finding is important as depression and anxiety are prevalent, costly and vastly undertreated disorders.

Abstract Body

Introduction

Adaptive Treatment Strategies warns therapists of patients at risk of treatment failure to prompt an adaption of the intervention. Internet-delivered Cognitive Behavioural Therapy (ICBT) collects a wide range of data before and during treatment and can quickly be adapted by adjusting the level of therapist support.

Objectives

To evaluate how accurate machine learning algorithms can predict a single patient’s final outcome and evaluate the opportunities for using them within an Adaptive Treatment Strategy.

Methods

Over 6000 patients at the Internet Psychiatry Clinic in Stockholm receiving ICBT for major depression, panic disorder or social anxiety disorder composed a training data set for eight different machine learning methods (e.g. k-Nearest Neighbour, random forest, and multilayer perceptrons). Symptom measures, messages between therapist and patient, homework reports, and other data from baseline to treatment week four was used to predict treatment success (either 50% reduction or under clinical cut-off) for each primary symptom outcome.

Results

The Balanced Accuracy for predicting failure/success always were significantly better than chance, varied between 56% and 77% and outperformed the predictive precision in a previous Adaptive Treatment Strategy trial. Predictive power increased when data from treatment weeks was cumulatively added to baseline data.

Conclusions

The machine learning algorithms outperformed a predictive algorithm previously used in a successful Adaptive Treatment Strategy, even though the latter also received input from a therapist. The next steps are to visualize what factors contributes most to a specific patient’s prediction and to enhance predictive power even further by so called Ensemble Learning.