Virtual Reality and Artificial Intelligence in Treatment for Patients With Psychosis

Introduction Virtual Reality (VR) immerses people in digitally simulated environments that can provide exciting interactive experiences and is thus well established in the entertainment industry. VR is however not just about fun and games; therapeutic use of VR is becoming increasingly popular for a broad range of mental health conditions such as anxiety disorders, PTSD, phobias and chronic pain. Although the concept of VR as a treatment modality has gained momentum and (better) use of technology is an explicitly stated goal in health care strategies and guidelines, the use of VR therapy in severe mental illness is much less developed and has received less attention. The important question therefore arises whether VR therapy actually improves clinical and functional outcome in psychosis and whether it is cost-effective. The RecoVRy study will answer these questions in the first randomized controlled trial (RCT) combining virtual reality with state-of-the-art Artificial Intelligence (AI). A strong multidisciplinary team of researchers and clinicians (psychiatry, clinical psychology, health economy, digital health, and occupational therapy) will run this RCT in close collaboration with a world-renowned expert on VR therapy in psychosis.

Schizophrenia and psychotic disorders Psychotic disorders are rare, with a lifetime prevalence close to 1 %. Still, they impose a major burden of illness in terms of both individual suffering and societal cost. They are severe and often chronic mental disorders characterized by a heterogenic set of positive and negative psychotic symptoms, neurocognitive impairment, and functional decline. Total costs associated with schizophrenia spectrum disorders in Norway are estimated to approximately 10.2 billion NOK per year making it the 5th most expensive health condition, with indirect cost due to productivity loss accounting for as much as 30 % and only between 5 % - 10 % having paid employment. Despite increased etiological understanding and refined treatment approaches, today's standard psychosis treatment does not prevent a large proportion of individuals with schizophrenia becoming isolated and inactive, which in turn negatively affects both physical and mental health. Individuals with schizophrenia are estimated to lose 13 - 15 years of potential life years. Reduced quality of life, social withdrawal and inactivity may largely be explained by avoidance due to symptoms of anxiety. As many as two-thirds of patients with schizophrenia have levels of anxious avoidance equivalent to those in individuals with agoraphobia. Symptoms of anxiety emerge for different reasons; hallucinations, fear of rejection, harm or social humiliation, paranoia, social inexperience, and risk of stigma, and may contribute significantly to the functional impairment observed in schizophrenia. Alleviating anxiety symptoms may thus be key to help improve functional outcomes and tackle the serious issues of social withdrawal and inactivity in individuals with schizophrenia.

Whilst the effect of antipsychotic medication in reducing positive psychotic symptoms and risk of relapse is well-documented, non-response and poor adherence to antipsychotic medication is frequent. Furthermore, negative symptoms are often associated with a limited response to pharmacotherapy. Cognitive Behavioral Therapy for psychosis (CBTp) is an established psychotherapeutic intervention recommended in several international and national guidelines. Meta-analyses have reported small effects of CBTp on psychotic symptoms, and there is considerable room for improvement. It has been suggested that an increased focus on the behavioral component of CBT, using repeated behavioral experiments, behavioral activation and exposure, may increase the effect of CBTp. Virtual Reality facilitates exactly that; a greater focus on behavioural aspects of CBTp.

System-level factors also challenge the delivery of appropriate psychological treatments such as CBTp. Mental health services worldwide are faced with increased patient contact volume and clinician caseload size. Implementing CBTp requires resources often not present in the mental health system. It warrants therapists' education, a certain training volume, adherence to treatment and above all exposure and learning in real-world situations. Typically, clinicians have little to no time to accompany patients in sessions outside the clinic. This imbalance between demand and supply of mental health services underlines the need for an intervention directly addressing these issues, such as the RecoVRy study to bridge this gap.

Virtual Reality Exposure Therapy (VRET) VRET is a promising approach to ensure high-quality therapy and possibly increase effects - optimizing the therapist's time in a mental health system with limited resources. VRET utilizes modern technology to enable a more refined approach to exposure therapy, also within the context of CBTp. VR refers to a computer-generated environment with different scenes and objects that appear real, making the user feel that they are immersed in and can interact with their surroundings. In a clinical setting, therapist and client may use VRET to address difficult and complex situations in a highly controlled environment, in real time, eliciting responses equivalent to those in a real situation. The objective is to habituate learned fear responses or disprove catastrophic beliefs held by the client. VR provides ample opportunity to repeatedly challenge fears and beliefs first-hand, instead of addressing them through conversation in a classic therapeutic setting (later). VRET has been found to be a well-tolerated, safe treatment method without adverse effects in individuals with psychosis. Studies examining the effect of VR therapy on social skills and VR augmented CBT in individuals with social anxiety and paranoid thoughts indicate higher motivational levels, symptom and social skills improvements and reduced social avoidance. Combined with CBT, VR also reduced social anxiety and paranoia. Another potential benefit of VRET in psychosis treatment is the possibility of easing patients into situations they would otherwise find too frightening and consequently avoid In the large UK based RCT gameChange, VR therapy led to significant reductions in distress and anxious avoidance in and of real-world situations in patients with psychosis.

RecoVRy - a new and improved Artificial Intelligence powered VR application. Based on experiences from a pilot study and important user guidance, we have developed a Norwegian VR application, RecoVRy, for individuals with psychosis experiencing anxiety in everyday social situations. Scenarios include public transportation, a waiting room, a job interview, giving a presentation in a work/educational setting, meeting with a case manager at the welfare services (NAV) and visiting a cafe. Clients are guided through scenarios by integrated instructions (orally or written) and gamification elements introducing psychological principles and aims of the treatment throughout each session. In the different scenarios, tasks are carried out, such as ordering a coffee or interacting with other passengers on the bus. RecoVRy is designed to be easy to use, engaging for patients and clinicians and delivered with the latest consumer equipment. Additionally, RecoVRy combines VR with state of the art Artificial Intelligence (AI) to further increase immersion and potentially strengthen effects of behavioural activation and exposure. In most VR applications, including gameChange, patients cannot interact with avatars (e.g. customers in a shop or other passengers on a bus). Regardless of what the patient does or says during session, the avatars will provide standard comments and replies, not necessarily fitting the situation. Experiences from the pilot study indicated a strong need for patients to be able to interact with avatars across the scenarios. AI elements thus contribute to move the VR field forward and have been included in the RecoVRy application. No previous large studies have described or examined the combined usage of these two emerging technologies in psychosis treatment. Each scenario has several difficulty levels and is designed to activate fearful cognitions, enabling patients to explore fears and paranoid thoughts, and trying new behaviors while at the same time limit the use of defence mechanisms and avoidance. Events take place that challenge patients to drop safety-seeking behaviors they normally use, with the ultimate goal of helping them adopt new behaviors to use when feeling anxious and consequently feel more confident in everyday situations. Each session lasts approximately 30 minutes. RecoVRy entails CBT components. Consequently, it is not required for CBT trained staff to deliver the therapy. RecoVRy represents a completely new way of delivering psychosis treatment and has never been tested in Norway, that is, the new AI powered scenarios have been tested in a small clinical pilot for acceptability testing and technical verification (N = 5). In the current study, the aim is to determine the effect on real-world functioning, psychiatric symptoms and quality of life in individuals with psychosis. Further, the investigators will establish barriers and facilitators with regard to the implementation of VR therapy in clinical practice through a qualitative sub-study in patients, clinicians and health care leaders/stakeholders. A health economic evaluation of RecoVRy will also be carried out.

This study will answer whether using AI powered VR in psychosis treatment results in improved health and functional outcomes, and whether this technology increases access to augmented CBT in patients with schizophrenia by adding specific components of CBTp. RecoVRy may shorten the length of treatment, saving time and cost. Findings from this study will be relevant for public and private mental health services in Norway and internationally on the implementation of technology-based interventions in psychosis RecoVRy will create potential for improvement of existing practise and personalized treatment. CBTp, including central components such as behavioral experiments and exposure, is well-established. However, far from every mental health professional has CBT training, and behavioral experiments and exposure are often too time-consuming to be implemented in clinical practice. Consequently, many patients do not receive CBT at all, or when receiving CBT, are not given the opportunity to expose themselves to situations causing anxiety under clinical guidance. The use of RecoVRy can help overcome these challenges and may increase treatment effects by using AI to further tailor exposure. It can be used both as stand-alone treatment and as an add-on to CBTp, ensuring treatment access for more people and the ability to provide safe and effective exposure and behavioral tests under clinical guidance. RecoVRy thus touches upon the prioritized thematic area of new ways of working and several clinical groups and personnel can deliver VR treatment. Moreover, studies have shown that learning occurring through VRET transfers to the real-world, i.e. what you overcome in VR, you overcome in real life, which is not always the case with classic, dialogue-based CBTp.

RecoVRy will fill knowledge gaps concerning treatment of anxiety in psychosis. Recent reviews and meta-analyses call for further development and customization of CBT in terms of personalized psychotherapy. Actively entering anxiety provoking situations and practicing more adapting ways of thinking and behaving is challenging in face-to-face therapy, but relatively easy in VRET. Thus far, only a handful of studies, with promising results, have examined the impact of VRET on severe mental disorders. None has examined the effect of VRET combined with AI. Experience from the computer game industry suggests that VR-assisted therapies can be much more enjoyable than traditional treatments, ensuring good take-up and continuous "use" of the therapy. Further research is thus warranted. The current RCT aims to explore if the positive outcomes from using VRET can be repeated in individuals with psychosis using a Norwegian application and whether this treatment is cost effective. Evidence from this study will be of scientific interest nationally and internationally. RecoVRy will create societal benefits. Based on the assumption that a substantial part of the social avoidance and inactivity in schizophrenia arises from anxiety, using AI powered VR as a new medium for therapy may be a good first-step for individuals who have avoided mental health care in the past. It is well-known from the field of social anxiety disorder that appropriate treatment may produce superior results, provided the treatment is actually offered. RecoVRy may increase treatment adherence, as it might be more appealing than traditional methods, encouraging patients to continue therapy. It ensures consistency - the same treatment program is offered to all users, with the possibility of tailoring according to individual needs. RecoVRy can help shorten the length of treatment, will cost less than traditional therapy and can help therapists provide treatment to individuals at their convenience. Most importantly, social inclusion and self-governance are of high societal and individual value. Overcoming symptoms of anxiety may directly translate to greater social, educational and occupational participation - reducing costs and creating benefits.

Knowledge from RecoVRy can easily be generalized to other populations and health sectors. RecoVRy is thoroughly described and will be manualized when completed. The knowledge gained from this study, combined with already existing material based on VR for psychosis, will make it easy to generalize the knowledge and practicalities to other patient populations, age groups and health sectors. It will be particularly relevant for disorders in which social anxiety and withdrawal strongly contribute to the clinical consequences, such as phobias, mood disorders and other anxiety disorders.

1. Hypotheses, aims and objectives The overall aim of this project is to provide novel information about Virtual Reality therapy by carrying out a randomized controlled trial of AI powered VR treatment in psychosis, the RecoVRy application.

• The primary objective is to examine short- and long-term effects of VR therapy on anxiety and avoidance in patients with psychosis, including mediators and moderators
• The secondary objective is to examine short- and long-term effects of VR therapy on positive and negative psychotic symptoms, depressive symptoms, worry, paranoia, insight, quality of life, social and functional activity, including mediators and moderators
• In addition, a health economic evaluation will be conducted and a qualitative approach will answer research questions regarding experiences with receiving, providing and implementing VR therapy (patients, clinicians, health care leaders) and barriers and facilitators to implementation of VR therapy in clinical practice.

Our hypothesis is that RecoVRy combined with Treatment as Usual (TAU) is superior to VR gaming and TAU and leads to post treatment improvements in anxiety, avoidance, psychotic symptoms, depression, worry, paranoia, insight, quality of life and social/functional activity. A working hypothesis is that RecoVRy is superior to the control group in terms of cost-effectiveness.

Primary and secondary endpoints The primary endpoint will be the Cognitions and Defence Behaviours Questionnaire (CDBQ), a clinically meaningful measure assessing worries, avoidance and ability to manage different social situations. Analyses will include all participants for whom data are available and according to the group participants are randomly assigned to, regardless of any deviation from protocol. Analyses will be carried out after the last follow-up assessment is completed with no interim analyses and CDBQ will be assessed at all time points.

The secondary endpoints; i) Changes in mean scores of anxiety, positive and negative psychotic symptoms, work and social adjustment, quality of life, social functioning, depressive symptoms, worry, paranoia, insight, hope and neurocognition in both groups ii) group differences in health care utilization/health economy Safety outcomes and therapy satisfaction will be assessed using the Simulator Sickness Questionnaire and The Therapy Satisfaction Scale at post assessment The RecoVRy intervention Patients will be electronically randomized to either RecoVRy + TAU or TAU. The intervention group will receive eight VR sessions of 30 minutes. Three sessions are considered minimum dose of therapy. A mental health professional (VR therapist) will be present when VR treatment is provided and will help put on the VR headset, start the program and encourage the patient to apply the learning from RecoVRy to real-world challenges by setting up homework tasks to be carried out between VR sessions. The application will run through Meta Quest 2/3®. RecoVRy is designed as a repeated behavioural procedure in which the main focus is for participants to learn that they are safer than assumed. Participants can proceed at their own pace, allowing for more or less than 8 (+/- 2) sessions of VR treatment. Scenarios and levels may be repeated if necessary.

Statistical analysis Statistical analysis will be carried out according to the intention-to-treat (ITT) principle. The primary and secondary outcomes will be analyzed using linear mixed-effects regression to model post-test and follow-up assessments concurrently. In addition, structural equation models and linear regression will be applied to examine potential mediators and moderators with regard to VR therapy. Statistical power estimates:100 participants will be enrolled, 50 in each group. This sample size takes into consideration an attrition rate of 10% and provides 80% power to detect clinically significant differences in the CDBQ from randomization to 8 weeks at a two-sided 5% significance level.

1. User involvement User involvement is planned according to the guidelines for user participation in research in the specialised healthcare service. User representatives have been consulted and involved from the early planning phase. One user representative, Joachim Vogt, is regularly consulted (clinic internal innovation funding) and has given feedback with regard to the program, assessments and implementation. He has also participated in the formation of the study and given feedback regarding the general structure and questions from a user point of view.
2. Ethical considerations Patients will receive both written and oral information about the study before they are asked to give their written consent to participate. All data will be stored according to approval. The VR treatment will be conducted by trained VR therapists, and the therapies will be supervised throughout the study period. The patients who do not wish to participate in the study will receive treatment as usual at the respective sites. Participants are informed that they can withdraw from the study whenever they want to, and that this will have no consequences for their further treatment at the in- and outpatient clinics. The research coordinators will keep a protocol for collection and reporting of adverse events.