Consistent with the design of the first version of SloMo [16,17] the Double Diamond [8], a human-centered design framework, was used to optimize SloMo for implementation, using inclusive design principles [9]. The Double Diamond is the most widely used co-design framework in the United Kingdom. This framework was selected because of its alignment with inclusive design principles and because it was used to develop the previous release of SloMo [16,17].

This work was undertaken as a quality improvement project within South London and Maudsley NHS Foundation Trust and, according to institutional policy, did not require review or approval by a research ethics board. Participants provided verbal consent to participate in user-testing sessions and focus groups. Participation was voluntary, and all data were collected anonymously. No identifiable personal or health information was recorded, stored, or analyzed. The project was carried out in compliance with relevant ethical standards for quality improvement initiatives. Participants received gift vouchers for their contributions in line with National Institute for Health and Care Research payment rates [27].

SloMo (release 1; R1) was self-certified as a class I medical device. We anticipate SloMo release 2 (R2) to be a Class I medical device under Rule 12 (All other active devices; MEDDEV Guideline 2.4/1 Rev 9), as the criteria for Rules 9, 10, and 11 for active devices are not met. The software was programmed by Bitam Ltd, which met ISO (International Organization for Standardization) 13485 and ISO14971 standards (ISO13485 is an international standard for medical device quality management systems; ISO14971 is an international standard for risk management of medical devices). The SloMo therapy platform architecture is compliant with ISO/IEC 27001:2022, 27001:201, ISO/IEC 27017:2015 and 27018:201.

To address the risk inherent in participatory design that the most willing, able, and vocal users are more likely to be involved, neglecting the needs of underserved groups, we purposively sampled people from a wide range of backgrounds (ie, across diverse gender, age, ethnicity, technology confidence, and attitudes to therapy). The group was representative of service users across the 3 NHS Trusts, with respect to ethnicity and age (Table 1). Confidence in using technology was self-reported by service user consultants at the start of testing sessions on a scale of “0” (not confident at all) to “100” (very confident). A wide range of technology confidence was self-reported, skewed towards people with higher levels. This is in line with recent literature indicating digital literacy is improving amongst people with psychosis [28].

Therapist patient and public involvement (PPI) consultants currently working with people with psychosis were purposively sampled in relation to gender, age, ethnicity, and experience of delivering psychological interventions for psychosis. Ten therapist PPI consultants participated in the deliver phase. The mean age was 34.9 years (range 26-54) and comprised 7 women and 3 men. One therapist identified as Bengali, 2 as Black British-African, 2 as Black British-Caribbean, 4 as White British, and 1 as White (other). Seven therapists were clinical psychologists or counseling psychologists, one was a high-intensity CBT therapist, two were trainee clinical psychologists, and one was an assistant psychologist. The mean technology confidence score was 86.2 out of 100 (range 70 to 100).

Transdisciplinary co-design workshops (referred to as “workshops” here forward) were central to the redesign of the new version of SloMo. Our project team included experts by experience (EBEs), designers, clinicians, academics, software developers, video producers, and illustrators (Multimedia Appendix 1 contains further details of the co-design team). Workshops were initially held in person (to support group cohesion), then switched to remote (to facilitate engagement of people from geographically diverse locations). Attendees were informed of the workshop’s focus in advance of the meeting, which was prespecified according to the phase of development. Sessions were recorded so that the discussion could be reviewed, and members who were unable to attend could review and offer feedback via email. A PPI management team oversaw the project.

The Double Diamond is organized into 4 convergent and divergent phases: discover (understand the problem; divergent), define (reframe the problem and develop a design brief; convergent), develop (develop potential solutions; divergent), and deliver (test out solutions at a small scale and iterate; convergent). It should be noted that divergent phases were necessarily constrained by the previous version of SloMo and its clinical evaluation, and so we adopted a “waterfall-agile” approach [16]. Figure 1 provides an overview of the Double Diamond phases used to optimize SloMo for implementation, which are explained in more detail below.

The aim of the discover phase was to develop an understanding of possible implementation barriers for the SloMo software. This first involved reviewing the outputs of the SloMo RCT [15,18-20] in the transdisciplinary co-design workshops, to explore key technology problem areas that could potentially be targeted for the next release of SloMo. In addition, the designers conducted qualitative interviews with 2 EBEs who had received the first release of SloMo. Interviews focused on exploring what was required from a psychological therapy for psychosis, building on insights from Hardy et al [17], as well as exploring how SloMo could be improved.

The define phase synthesized the insights gathered in the transdisciplinary workshops. The EBEs who provided feedback during the discover phase also took part in the next round of workshops to assist in synthesis. The information from the discover phase was sent to co-design group members to review ahead of time to allow for the opportunity to familiarize and reflect on the materials. Workshops involved reviewing the insights as a group, noting patterns, and sorting into themes, using a group consensus approach. Multiple workshops were run until a clear and workable design brief was developed, in line with identified user needs.

The develop phase aimed to generate potential design solutions that helped to address the redefined user needs outlined in the preceding phase. Guided by clinician and EBE recommendations, the designers initially explored content relevant to SloMo (ie, existing apps available on the market, including for mental health, as well as apps in parallel markets, such as meditation, nutrition, and habit-building). The designers fed back their design solutions in co-design workshops, and group consensus determined which ideas were taken forward for iterative development.

Low-fidelity digital prototypes of initial design concepts were developed (Multimedia Appendix 2). Preferred concepts, selected through workshops, were taken for one-to-one user testing with service user PPI consultants. Data from the user testing sessions informed the iterative development of design solutions, leading to the development of higher fidelity digital prototypes of adopted concepts. Prototypes were further iteratively tested and developed through one-to-one service user testing, focus groups with service user and clinician consultants, and workshops. Once a prototype had been validated, it was added to the optimized SloMo minimum viable product (MVP) software, which was taken forward to the deliver phase.

The convergent deliver phase aimed to validate the user experience of the optimized version of SloMo. Given the redesign of SloMo focused on 2 software components (a web-based therapy platform and native mobile app), the deliver phase aimed to validate both through consultation with the applicable user groups. For the web-based therapy platform, therapist PPI consultants were chosen as this would be used by therapists to augment the face-to-face therapy sessions they deliver. Service user PPI consultants were chosen for mobile app feedback, as they were the main intended users of the app.

Concurrent think aloud testing sessions were carried out with therapist PPI consultants. This is a form of usability testing where participants are asked to verbalize their thoughts in real time as they interact with a user interface (UI; ie, modules on the SloMo web-based therapy platform) [29]. Following this, additional unstructured qualitative feedback was gathered. A similar approach was taken for the mobile app. Service user PPI consultants who were involved in the develop phase took part in think-aloud testing sessions. Service users were asked to complete different core functionalities on the app (eg, adding a new worry, slowing down a worry, viewing session summary cards) whilst thinking aloud. Interactions were observed, and qualitative feedback was elicited. Following this, a user experience survey (UES) [19] was completed. This is a 10-item questionnaire (adapted from Lobban et al [30]) with subscales relating to enjoyment, usability, and usefulness. Each item is rated from 0 to 10 and summed for the subscale total scores, which were calculated as percentages, with higher percentages indicating a more positive experience. Following user experience validation, an MVP of SloMo, optimized for implementation, was finalized for release to Apple and Android app stores for piloting clinical use.

The key insights and outputs from each phase of the Double Diamond are presented in Table 2 and expanded on below. Additional details on participants, methods, and outputs can be found in Multimedia Appendix 3 [15,17-20].

Key implementation barriers in the technology domain for the first software version were identified through review of the SloMo RCT outputs and interviews with EBEs. This included reporting of poorer software usability by men and people with lower levels of technology confidence. In addition, qualitative data revealed service users disliked carrying 2 mobile handsets (the first release of SloMo was only available on a study handset). Therapists reported difficulties in pacing the sessions, as there were no signifiers indicating how much session content remained, which aligned with service user feedback indicating progress was at times too slow. Therapists and service users reported finding the content of psychoeducational vignettes limited and instead used SloMo for a greater range of paranoia-related worries than was covered in the software. Other barriers included a lack of clarity around when data was being transferred between devices (ie, the web-based therapy platform and mobile app), an overemphasis on worries instead of safety on the app home screen, difficulties in generating personalized goals and safer thoughts, a lack of diversity across psychoeducational vignettes, and forgetting content covered in therapy sessions once these had ended. Outputs at the end of the discover phase included a list of technology-related implementation barriers. These barriers were reviewed during transdisciplinary workshops and taken forward to the define phase.

The define phase reaffirmed the importance of SloMo’s design brief of users needing a form of CBTp that was usable, trustworthy, enjoyable, personalized, normalizing, and memorable [17]. The problems identified during the discover phase were mapped onto these domains, indicating these were areas that could be targeted in the next iteration to further deliver on this design brief and enhance user experience.

The issues outlined in the discover phase were reframed as user needs through co-design workshops. Service users needed a visually appealing UI, an intuitive mobile app navigation, and convenient access to support usability. To foster a sense of trust, service users required greater agency over data sharing to and from the web-based therapy platform and mobile app. Service users needed to feel psychologically supported when opening the mobile app, for SloMo to feel like a positive experience, and to feel represented throughout. There was a need for learning ways of managing a wider range of paranoia-related worries, and for the SloMo journey to continue once face-to-face sessions had ceased. For therapists, it was important that they were provided feedback on session progress so they could pace sessions efficiently. There was a need for therapists to be able to use the web-based therapy platform flexibly, allowing them to tailor sessions to the client. The outputs at the end of the define phase included a design brief of the solution, which specified the user needs to be addressed in relation to the original design definition, which informed the subsequent development phase.

Thirty-two 1:1 service user testing sessions and 3 service user focus groups (n=9) were carried out. Initial user testing focused on design concepts, and later stages focused on interactive digital prototypes. Initial design concepts taken to 1:1 user testing sessions included (1) aesthetics of bubbles used to convey worry and safer thoughts, (2) design of the avatar representing the SloMo user on their journey, (3) SloMo character illustrations, and (4) aesthetics of physical tip cards used as a memory aid. Other concepts (eg, lived experience vignettes, UI changes, data transfer updates) were iteratively developed in co-design workshops. Multimedia Appendix 4 displays the updated concepts for the new version of SloMo compared with the previous one used in the RCT.

High-fidelity interactive digital prototypes developed and taken out for individual user testing included (1) user interactions with thought bubbles, (2) user interactions for building a customizable avatar, (3) character illustrations featured in audio-visual psychoeducational vignettes, and (4) the content of newly developed psychoeducational vignettes. User testing feedback in both the concept and prototype development phases was discussed in co-design workshops and iteratively developed until there was a group consensus that the prototype should be adopted. Figures 2 to 4 provide an overview of the development of specific SloMo features from the discover to deliver phase. Multimedia Appendix 5 provides additional data collected during the develop phase.

Outputs at the end of the develop phase included iteratively developed digital design solutions addressing the user needs and design brief from the define phase. In addition, early-stage demonstration and interaction testing of design solutions with user testing consultants provided preliminary validation of specific design solutions. These were then developed into the SloMo (R2) MVP during the deliver phase.

Table 3 presents quotes from the therapist PPI consultants during the think-aloud testing sessions. Overall, the web-based therapy platform received positive feedback. Therapists appreciated the platform’s simplicity and visual appeal, believing it would enhance the enjoyment and engagement of sessions. They noted therapeutic benefits, such as providing more structure to sessions and making therapy feel more personalized. Additionally, therapists found the platform engaging and flexible, allowing for adaptation based on various therapeutic needs. The addition of progress bars was positively commented on as a helpful feature for therapists to manage the session time more effectively.

During the think-aloud sessions, several areas for improvement were identified. Some therapists expressed hesitation about navigating back and forth due to concerns that personalized content entered into the platform might be lost. Others suggested that enjoyment could be further enhanced by incorporating more gamification elements, such as badges or achievements, to boost engagement. Additionally, there were recommendations to increase inclusivity by adding different languages, enabling access for users whose first language is not English.

Eleven service user PPI consultants who had previously taken part in user testing feedback sessions in the development develop phase took part in the concurrent think-aloud testing sessions for the deliver phase. Think-aloud consultants were purposively sampled to be representative of the psychosis population across the respective areas of the 3 trusts. The sample included 7 women and 4 men, whose ages ranged from 25 to 60 years, and 4 consultants identified as Black British, 1 as South American, and the remaining 6 as White British.

The UES findings for each subscale are presented in Table 4. The ratings suggest a good user experience of the SloMo demonstration, with a total UES mean of 83.4%. Notably, the mean score for enjoyment was around 10% lower than the domains of usability and usefulness, as well as showing a greater range of scores. This suggested that improved enjoyment may be an area requiring further development.

Table 5 displays qualitative feedback captured during think-aloud user testing sessions with service user PPI consultants. The feedback aligned with the UES ratings, indicating areas across the 3 domains of user experience that were well received. Interestingly, there were suggestions around how enjoyment could be enhanced, possibly providing insight into lower UES scores for this domain. For instance, haptic feedback was seen as a feature that could boost enjoyment, and improving touch screen interactions could further enhance usability. Outputs at the end of the develop phase included an MVP of SloMo (R2) that had received early-stage user experience validation from service users and therapist consultants. The user experience of the software will be further evaluated in an ongoing implementation-effectiveness study. Figure 5 [16,31] illustrates the prototype delivered from the develop phase.

This study optimized the SloMo therapy software for implementation in routine NHS care, through reducing technology complexity and boosting user experience [24,25]. An iHCD was used, involving transdisciplinary co-design workshops, focus groups, one-to-one user testing, and think-aloud testing sessions. From this research, we have produced an optimized MVP of SloMo, which is now being tested in the SloMo2 hybrid implementation-effectiveness study (ClinicalTrials.gov identifier: NCT06568081). User testing indicated that users need a form of CBTp that is usable, trustworthy, enjoyable, personalized, normalizing, and memorable, in line with earlier work from our team [17]. Design targets for improving the SloMo software were identified to deliver this design brief and address user needs. For reducing complexity, key targets included simplified app interfaces to highlight core functionalities, improved navigation, and, for boosting user experience, this included providing users with greater agency over data, and improving the representation of protected characteristics, including their intersectionality. The user experience of this optimized version of SloMo may be improved relative to the first version that was tested in the SloMo RCT, indicated by higher UES scores compared with those reported by Hardy et al [19].

A distinctive aspect of our product development approach is the integration of lived experience involvement in all stages of the design work. Traditional approaches to intervention development in clinical research often follow a convergent design process using existing theory and evidence to determine intervention content and structure. This approach can miss novel opportunities to improve the user experience of therapies. In contrast, we adopted a “waterfall-agile” approach to facilitate the co-design of SloMo as an evidence-based digital therapeutic [16]. This allowed for agility in iterating specific features (eg, UI, psychoeducational vignettes) while maintaining fidelity to the previous UI and therapy protocol [18]. To support this, we embraced divergence, revisiting and challenging earlier design assumptions in line with iHCD principles. An example of this was the service user feedback that “unusual experiences” could be perceived as stigmatizing, despite this wording being widely used in CBTp [32] and the identification of a term describing these types of experiences, which felt more relevant and acceptable—“when the world doesn’t feel right.” Despite this divergent approach, we were necessarily constrained in certain respects. The need to build on a preexisting evidence base meant that some design decisions were necessarily limited by prior research findings and the parameters of the therapeutic model. For instance, some users suggested moving away from visualizing thoughts and thinking habits as bubbles, which was not possible, as this is a core component of SloMo and its previous clinical evaluation.

Resource constraints restricted our ability to optimally develop this new version of SloMo. This included time pressure, as there was a need to produce an MVP in line with our implementation-effectiveness study milestones. Another constraint was workforce access, such as access to PPI consultants (therapists and service users), designers, and software developers. These limitations meant that pragmatic decisions had to be made for the redesign, for instance, regarding the types of software changes that could be made and what features were selected for user testing. Whilst decisions were made collaboratively within the transdisciplinary co-design workshops, it is possible that different design decisions may have been made with additional resources. These constraints may be viewed as a barrier to reducing complexity in the technology domain of the NASSS framework [24,25]. Ideally, future projects could be based in “living labs,” whereby lived experience, clinical, technological, industry, regulatory, and research expertise is brought together in an organizational structure, with coproduction underpinning the co-design of digital health [33].

A limitation of the PPI consultant validation is that the service user and therapist experience of the optimized version of SloMo was based on a relatively brief exposure to the software, and so may not be directly comparable to the user experience findings from the SloMo RCT. Further, the PPI consultants involved in the delivery phase had been involved in the SloMo design work for an extended period and may have been less independent than the RCT sample. This may limit the conclusions we can draw from this early-stage validation. However, more rigorous data will be provided in our ongoing implementation study, examining SloMo when delivered by frontline therapists.

Validation of the SloMo (R2) MVP with service user consultants found enjoyment was rated lower than other domains on the UES [17], indicating an aspect of user experience that warrants attention in future development. Qualitative feedback highlighted specific areas for improvement, such as incorporating haptic feedback and improving user interaction. The digital therapy engagement literature could inform additional design strategies [34]. Nonetheless, lower enjoyment scores may reflect comparisons with nontherapeutic commercial apps, where higher levels of enjoyment are likely expected. It will be important to monitor this throughout the implementation-effectiveness study, and iterate on the software as needed; for example, monitoring engagement may act as a proxy for enjoyment.

This study demonstrates the value of using iHCD to optimize DMHIs for implementation in routine mental health care. By adopting an iterative, transdisciplinary approach that prioritized collaboration with service users, therapists, and technical experts, the SloMo redesign addressed critical usability barriers, fostered inclusivity, and enhanced the therapy’s ability to meet the diverse needs of its target population. This methodology offers a valuable model for future mental health innovation, where the needs of diverse user groups are addressed, and therapeutic technologies are designed to be impactful and sustainable in routine care settings. We are now evaluating the implementation and effectiveness of the optimized version of SloMo when delivered by routine care therapists. We envisage SloMo will continue to be iterated based on learning from this work, in line with the principles of implementation science [35].