Engagement and attrition in digital mental health: current challenges and potential solutions
數位心理健康的參與和流失：當前挑戰和潛在解決方案

Katharine A. Smith $^{1, 2, 3} ⊠$ , Thomas Ward $^{4, 5}$ , Sinéad Lambe $^{2, 6}$ , Edoardo G. Ostinelli $^{1, 2, 3}$ , Charlotte Blease $^{7}$ , Thomas Gant $^{4, 5}$ , Stefan M. Gold $^{8, 9, 10}$ , Emily A. Holmes $^{7}$ , Ivana Paccoud $^{11}$ , Anastasia Vinnikova $^{12}$ , Jochen Klucken $^{11, 13}$ , Peter J. Uhlhaas $^{14}$ , Carolina Garcia Sanchez $^{7}$ , Kate Haining $^{15}$ , Kerem Böge $^{8, 9}$ , Sofiia Lahutina $^{16, 17}$ , Luisa Tomelleri $^{18}$ , Sean Ryan $^{19}$ , John Torous $^{19, 20}$ & Andrea Cipriani $^{1, 2, 3, 20}$
凱薩琳·史密斯 $^{1, 2, 3} ⊠$ ，湯瑪斯·沃德 $^{4, 5}$ ， Sinéad Lambe $^{2, 6}$ ，愛德華多·奧斯蒂內利 $^{1, 2, 3}$ ，夏洛特·布萊特 $^{7}$ ，湯瑪斯·甘特 $^{4, 5}$ ，斯特凡·戈爾 $^{8, 9, 10}$ 德，艾米麗·福爾摩斯 $^{7}$ ，伊萬娜·帕庫德 $^{11}$ ，阿納斯塔西婭·文尼科娃 $^{12}$ ，約亨·克魯肯 $^{11, 13}$ ，彼得·烏哈斯 $^{14}$ ，卡羅琳娜·加西亞·桑切斯 $^{7}$ ，凱特·海寧 $^{15}$ ，凱雷姆·博格 $^{8, 9}$ ，索菲亞·拉胡蒂娜 $^{16, 17}$ ，路易莎·托梅勒里 $^{18}$ ，肖恩·瑞恩 $^{19}$ ，John Torous $^{19, 20}$ & Andrea Cipriani $^{1, 2, 3, 20}$

In digital mental health engagement rates are consistently low, which may limit its effects. Using an international multidisciplinary consensus method, including lived experience expertise and a systematic review, we identified three key challenges: (i) lack of agreed metrics for engagement; (ii) lack of evidence on how better engagement improves outcomes; (iii) lack of standards for user involvement. Three potential solutions encompassed: (i) standardisation of frameworks for reporting engagement metrics and optimal doses of digital tools, (ii) measuring engagement with more precise reporting of outcomes, including potential harms; (iii) defining standards of user involvement (including appropriate diversity, and clinician as well as user input). Digital interventions have real potential in meeting the shortfall in service provision for mental health, but this will require focus on high quality research studies of the underlying mechanisms of engagement and optimal outcomes. Our findings identify and highlight the next best steps in this process.
在數位心理健康中，參與率一直很低，這可能會限制其影響。使用國際多學科共識方法，包括生活經驗專業知識和系統評價，我們確定了三個關鍵挑戰：（i）缺乏公認的參與度指標;（ii）缺乏關於更好的參與如何改善結果的證據;（iii）缺乏使用者參與標準。三種可能的解決方案包括：（i）報告參與度指標和數位工具最佳劑量的框架標準化，（ii）通過更精確地報告結果（包括潛在危害）來衡量參與度;（iii）定義使用者參與的標準（包括適當的多樣性、臨床醫生和用戶的意見）。數字干預在填補心理健康服務提供的短缺方面具有真正的潛力，但這需要專注於對參與和最佳結果的潛在機制的高品質研究。我們的研究結果確定並強調了這一過程中的下一個最佳步驟。

Engagement is an essential element for any digital health tool

^{1}

. However, user engagement with digital mental health interventions (DMHIs) as assessed by usage data remains consistently low

^{2}

. This presents a fundamental challenge: how can tools designed to support mental health succeed if users do not actively use or engage with them?
參與是任何數位健康工具

^{1}

的基本要素。然而，根據使用數據評估的使用者對數位心理健康干預（DMHI）的參與度一直很低

^{2}

。這就提出了一個根本性的挑戰：如果使用者不積極使用或參與旨在支援心理健康的工具，它們如何取得成功？

Digital interventions provide an unprecedented opportunity to study the usage aspects of engagement; unlike traditional healthcare settings where engagement can be challenging to quantify, digital tools can generate real-time, automated usage data such as logins, pageviews, time used and even eye gaze that offer valuable insights into user behaviour

^{3}

. However, engagement involves more than this, as usage data
數字干預為研究參與度的使用方面提供了前所未有的機會;與傳統的醫療保健環境不同，在傳統的醫療保健環境中，參與度可能難以量化，而數位工具可以生成即時、自動化的使用數據，例如登錄、流覽量、使用時間，甚至眼神凝視，從而提供對用戶行為的寶貴見解

^{3}

。然而，參與度涉及的不僅僅是使用數據
alone does not capture the degree of investment of the participant in the activity

^{4}

. Thus, many digital health studies have also used qualitative methods such as questionnaires or semi-structured interviews to help capture some of the more nuanced and complex aspects of engagement, including satisfaction, acceptability and usability. Despite these efforts, there has been huge variability in the terminology applied to both usage and self-report measures

^{5, 6}

. Although several conceptual frameworks for engagement have been proposed

^{7}

, capturing the complex construct of meaningful engagement, including not only usage, but also its cognitive, emotional and behavioural dimensions, has proved challenging

^{8}

. The variation in definitions has also impeded meaningful comparisons of
單獨並不能捕捉參與者對活動的

^{4}

投入程度。因此，許多數位健康研究還使用了定性方法，例如問卷調查或半結構化訪談，以説明捕捉參與度的一些更細微和複雜的方面，包括滿意度、可接受性和可用性。儘管做出了這些努力，但應用於使用方式和自我報告措施

^{5, 6}

的術語仍然存在巨大差異。儘管已經提出了

^{7}

幾個參與的概念框架，但事實證明，捕捉有意義參與的複雜結構，不僅包括使用，還包括其認知、情感和行為維度，已被證明是具有挑戰性的

^{8}

。定義的變化也阻礙了對

metrics of engagement across studies

^{6}

and also across assessments of outcomes, because measured outcomes may underestimate the intervention treatment effects when engagement has been poor or variably reported

^{3}

.
跨研究

^{6}

以及跨結果評估的參與指標，因為當參與度差或報告

^{3}

可變時，測量的結果可能會低估干預治療效果。

Given these challenges in such a key area in digital health, we implemented a novel approach: we used a well-established consensus methodology, with international multidisciplinary expertise including lived experience, and incorporated a systematic search of the evidence to identify both the challenges and the potential solutions to guide the study and optimisation of effective engagement in DMHIs. We also considered how this might translate to improved outcomes for users.
鑒於數位健康這樣一個關鍵領域的這些挑戰，我們實施了一種新穎的方法：我們使用了一種成熟的共識方法，具有包括生活經驗在內的國際多學科專業知識，並結合了對證據的系統搜索，以確定挑戰和潛在的解決方案，以指導研究和優化 DMHI 的有效參與。我們還考慮了這如何轉化為改善用戶的結果。

Results 結果

The consensus group identified three broad areas of challenge in understanding engagement in DMHIs, which are summarised in Table 1.
共識小組確定了理解 DMHI 參與的三大挑戰領域，表 1 總結了這些挑戰。

The consensus group agreed on the lack of clarity across studies and the lack of universally agreed, standardised definitions of the metrics related to engagement

^{3, 8}

. Terms such as usage, adherence, engagement, and attrition are used to describe aspects of engagement, and empowerment to describe patient related outcomes of engagement, but they often have overlapping definitions and vary between individual studies. For example, ‘engagement’ is often equated directly to the frequency or duration of usage but their definitions vary between studies

^{3, 5, 9}

. In addition, there was consensus that studies often do not report the raw data needed to make comparisons between studies. For example, in a scoping review focussing on DMHIs for depression, only 59% (13/22) of studies reported usage statistics

^{10}

. Within usage statistics themselves, studies often do not report the original data but instead a measure of ‘adherence’, usually an assessment of compliance with a pre-specified metric of completion (e.g., a set number of modules). Some definitions of adherence are more nuanced and incorporate not only usage and intended use but also justification for how intended use was defined

^{11}

; however, definitions vary and often lack a clear rationale for their selection. Even in studies where usage or adherence are reported, this does not necessarily reflect engagement, which is a more complex concept involving not only usage but also cognitive, affective and motivational components

^{12}

. The individual concepts are related (for example, usage is a pre-requisite for engagement and adherence), but they differ in scope and complexity.
共識小組同意研究之間缺乏明確性，並且缺乏普遍同意的標準化定義與參與度

^{3, 8}

相關的指標。使用、依從性、參與度和流失率等術語用於描述參與度的各個方面，而授權用於描述與患者相關的參與度結果，但它們通常具有重疊的定義，並且因個別研究而異。例如，「參與」通常直接等同於使用的頻率或持續時間，但它們的定義因研究

^{3, 5, 9}

而異。此外，人們一致認為，研究通常不報告進行研究間比較所需的原始數據。例如，在一項側重於 DMHI 治療抑鬱症的範圍界定審查中，只有 59% （13/22）的研究報告了使用統計數據

^{10}

。在使用情況統計本身中，研究通常不報告原始數據，而是報告“依從性”的衡量標準，通常是對符合預先指定的完成指標（例如，一定數量的模組）的評估。依從性的一些定義更加微妙，不僅包括用途和預期用途，還包括如何定義

^{11}

預期用途的理由;然而，定義各不相同，並且通常缺乏明確的選擇理由。即使在報告了使用或依從性的研究中，這也不一定反映參與度，這是一個更複雜的概念，不僅涉及使用，還涉及認知、情感和動機成分

^{12}

。各個概念是相關的（例如，使用是參與和遵守的先決條件），但它們的範圍和複雜性不同。

Despite these challenges in definition, the consensus group agreed that engagement remains a key concept for study in DMHIs as this is an essential first step in effecting positive outcomes. Digital interventions provide a unique opportunity to investigate the mechanisms of engagement through their ability to automatically capture detailed data on usage patterns, user interactions, and associated outcomes

^{3}

and identify interventions that can maximize effective engagement and positive outcomes.
儘管在定義上存在這些挑戰，但共識小組一致認為，參與仍然是 DMHI 研究的關鍵概念，因為這是實現積極結果的重要第一步。數字干預提供了一個獨特的機會，通過它們能夠自動捕獲有關使用模式、使用者交互和相關結果

^{3}

的詳細數據，並確定可以最大限度地提高有效參與和積極結果的干預措施，從而研究參與機制。

Does better engagement improve clinical outcomes?
更好的參與度會改善臨床結果嗎？

Engagement requires effort and has inherent limits for each individual. In DMHIs, this challenge is further compounded by the conditions they aim to address - many mental health disorders include low motivation and impaired concentration as core symptoms, potentially hindering sustained engagement in diverse and unpredictable ways. In contrast, digital approaches could have advantages for some specific mental health disorders. For example, virtual reality (VR) exposure therapy for anxiety might be expected to improve treatment retention. In fact, study results have been equivocal, with similar attrition rates between VR and in-vivo exposure treatments

^{13}

, although the data are difficult to assess as the original studies often did not report reasons for dropout and used now-obsolete technology.
參與需要努力，並且對每個人來說都有內在的限制。在 DMHI 中，他們旨在解決的條件進一步加劇了這一挑戰——許多心理健康障礙包括動力不足和注意力不集中作為核心癥狀，可能會以不同和不可預測的方式阻礙持續參與。相比之下，數位方法可能對某些特定的心理健康疾病具有優勢。例如，針對焦慮症的虛擬實境（VR）暴露療法可能有望提高治療保留率。事實上，研究結果是模棱兩可的，VR 和體內暴露治療

^{13}

之間的損耗率相似，儘管數據很難評估，因為原始研究通常沒有報告退出的原因並且使用了現已過時的技術。

In general, attrition has not been widely studied. Factors which have been identified as being involved in early engagement relate more strongly to perceived rather than objective need, and reasons such as forgetting, not finding time, or not finding the digital intervention useful have been associated with attrition

^{14}

. Additionally, attrition rates are likely to vary by mental health condition, treatment type and stage of illness. However, attrition, as well as a marker of loss of engagement, can also be a marker of treatment success. There are positive reasons for disengagement: the participant may have internalised their learning and be using this outside the digital space, or may be using other resources, or have achieved recovery

^{5, 14}

. Non-adherence or attrition may also reflect ‘e-attainment’-the discontinuation of engagement because personal goals have been met

^{15}

. In addition, there is no agreed approach for how to assess users who are nonengaged but stay in the study. Effective use patterns may differ from user to user

^{15}

. For example, Chien et al.

^{16}

identified 5 discrete subtypes of users based on engagement and found that the level of engagement was not always proportional to the observed clinical improvements

^{16}

.
一般來說，流失尚未得到廣泛研究。已被確定為參與早期參與的因素與感知需求而不是客觀需求的關係更緊密，而忘記、找不到時間或不認為數位干預有用等原因與流失有關

^{14}

。此外，流失率可能因心理健康情況、治療類型和疾病階段而異。然而，流失以及失去參與度的標誌也可能是治療成功的標誌。脫離參與有積極的原因：參與者可能已經內化了他們的學習並在數字空間之外使用它，或者可能正在使用其他資源，或者已經實現了恢復

^{5, 14}

。不依從或流失也可能反映「e-achieveinment」——因為個人目標已經實現

^{15}

而停止參與。此外，對於如何評估未參與但仍留在研究中的使用者，尚無公認的方法。有效的使用模式可能因用戶

^{15}

而異。例如，Chien 等人

^{16}

根據參與度確定了 5 個離散的用戶亞型，發現參與度並不總是與觀察到的臨床改善

^{16}

成正比。

Although it is widely accepted that engagement with digital interventions should be positively associated with improvements in mental health, this has been difficult to demonstrate robustly

^{17}

. Usage is often reported as an outcome in itself, but whilst some usage is needed, there has been little research on what may be the optimal or ‘target’ dose to achieve effective outcomes. For example, a systematic review of DMHIs suggested that greater usage may be correlated with improvements in mental health

^{18}

, but the interpretation was limited as this was measured differently in the different studies. In a review focussing on digital interventions in depression, a small number of studies (14) measured the relationship between usage metrics and outcomes, and of these 9 found an association between increased engagement and improved participant outcomes

^{8}

. In contrast, in some conditions, such as post-traumatic stress disorder (PTSD), ultra-brief treatments have been effective and acceptable

^{19, 20}

, challenging the idea that longer engagement or usage are always required. Whilst the usage metrics reported in some research studies may appear promising, this may not always translate to everyday or clinical settings. For example, a review of unguided e-mental health interventions showed that in research studies which proactively recruited users, the median programme usage rate was
儘管人們普遍認為參與數位干預應該與心理健康的改善呈正相關，但這很難有力地

^{17}

證明。使用量本身通常被報告為一種結果，但雖然需要一些使用，但很少有關於實現有效結果的最佳劑量或 “目標 ”劑量的研究。例如，一項對 DMHIs 的系統評價表明，更多的使用可能與心理健康的改善相關

^{18}

，但解釋是有限的，因為在不同研究中測量的程度不同。在一項專注於抑鬱症數位干預的綜述中，少量研究（14）測量了使用指標與結果之間的關係，其中 9 項研究發現參與度增加與參與者結果

^{8}

改善之間存在關聯。相比之下，在某些情況下，例如創傷后應激障礙（PTSD），超短期治療是有效且可接受的

^{19, 20}

，挑戰了總是需要更長時間參與或使用的想法。雖然一些研究中報告的使用指標可能看起來很有希望，但這可能並不總是轉化為日常或臨床環境。例如，對無指導性電子心理健康干預的審查表明，在主動招募使用者的研究中，程式使用率的中位數為

Table 1 | Areas of challenge in assessing engagement and adherence in digital studies in brain health (as identified by the consensus group)
表 1 |評估腦健康數位研究的參與度和依從性的挑戰領域（由共識小組確定）

Lack of universally agreed definitions of metrics related to engagement
缺乏與敬業度相關的指標的普遍認可的定義

Terms such as usage, engagement, adherence, attrition, and empowerment all have variable definitions in individual studies
使用、參與、依從性、損耗和賦權等術語在單個研究中都有不同的定義
Raw data are often not reported
原始數據通常不會被報告

Lack of evidence of how or whether improved engagement improves outcomes.
缺乏關於提高參與度如何或是否能改善結局的證據。

No clear evidence that increased engagement improves outcomes
沒有明確的證據表明提高參與度可以改善結果
No clear evidence of a dose effect of engagement or the optimal dose needed
沒有明確的證據表明參與的劑量效應或所需的最佳劑量
Difficulties of translating engagement from the research to real-world settings
將參與從研究轉化為現實世界環境的困難
No clear evidence on the exact mechanisms of engagement, individual effective use patterns and attrition
沒有關於參與的確切機制、個體有效使用模式和流失的明確證據
Interaction with mental health
與心理健康的互動
Attrition 磨損
Lack of adverse event reporting
缺乏不良事件報告

User involvement in developing and delivering digital health interventions
用戶參與開發和提供數位健康干預措施

User involvement can occur in different ways and levels of intensity and inclusivity
用戶參與可以以不同的方式以及強度和包容性水平進行
Reporting of user involvement is variable
用戶參與度的報告是可變的
User centred design may improve engagement and outcomes, but more direct evidence is needed
以使用者為中心的設計可能會提高參與度和結果，但需要更直接的證據
4.06 times higher than the subsequent real-world usage $^{21}$ . This may be related to the additional factors in trial settings (such as frequent human contact and extra assessments), which are less evident in real-world use. The mechanisms of engagement, including the essential elements or markers of effective engagement have not been clearly identified $^{3}$ . Engagement is a complex behaviour, usually starting with a prompt or an interest in adopting an intervention (for example, from a clinician, peer or social media) followed by initial use, and engagement. Disengagement and reengagement with the same or different intervention may also follow $^{22}$ . These stages vary in order and time course between individuals, between interventions and during the course of the intervention itself. Engagement strategies may only be effective at some stages-for example novelty may be helpful in initial signing up $^{14, 23}$ and habit formation may be more important in sustained use $^{24}$ .
比後續的實際使用方式 $^{21}$ 高 4.06 倍。這可能與試驗環境中的其他因素（例如頻繁的人工接觸和額外的評估）有關，這些因素在實際使用中不太明顯。參與機制，包括有效參與的基本要素或標誌尚未明確確定 $^{3}$ 。參與是一種複雜的行為，通常從提示或有興趣採用干預措施（例如，來自臨床醫生、同行或社交媒體）開始，然後是初始使用和參與。同樣或不同的干預也可能隨之而來 $^{22}$ 。這些階段在順序和時間進程上因個體、干預之間和干預本身而異。參與策略可能僅在某些階段有效——例如，新穎性可能在初始註冊中有所説明 $^{14, 23}$ ，而習慣的養成在持續使用 $^{24}$ 中可能更為重要。

User involvement in developing and delivering DMHIs
用戶參與開發和交付 DMHI

The consensus meeting agreed that involving end users in the design and delivery of digital interventions would be expected to enhance engagement and therefore, improve outcomes

^{25, 26}

. There are a variety of different approaches to involving the user (for example, co-production, co-design and human or user-centred design) and there has recently been increased interest in user-centred design in digital approaches

^{27, 28}

. However, usercentred design approaches are themselves often poorly defined and in practice user involvement in developing and delivering digital health interventions, including DMHIs is variable and often limited to the early and/or final stages of design development and delivery

^{25, 29, 30}

. In addition, reporting is variable, making assessments or comparisons of user involvement very challenging

^{25}

.
共識會議一致認為，讓最終用戶參與數位干預的設計和交付有望提高參與度，從而改善結果

^{25, 26}

。有多種不同的方法可以讓用戶參與進來（例如，共同生產、共同設計和以人或使用者為中心的設計），最近，人們對數位方法

^{27, 28}

中以使用者為中心的設計越來越感興趣。然而，以使用者為中心的設計方法本身往往定義不明確，在實踐中，使用者參與開發和提供數位健康干預措施（包括 DMHI）是可變的，並且通常僅限於設計開發和交付

^{25, 29, 30}

的早期和/或最後階段。此外，報告是可變的，這使得評估或比較用戶參與度非常具有挑戰性

^{25}

。

In digital mental health, there are only a few examples of true usercentred co-design. For example, a mapping review of studies in e-mental health interventions focussed on those where they identified user-centred design in their methodology. The papers were then analysed using the steps defined in the UK Design Council’s framework for innovation (https:// www.designcouncil.org.uk/our-work/news-opinion/double-diamond-universally-accepted-depiction-design-process/) and from this a variety of approaches were identified. Only 16 studies provided a definition of their chosen design approach and only 5 out of the 27 could be classified as using user-centred design

^{30}

.
在數位心理健康領域，真正以使用者為中心的協同設計的例子屈指可數。例如，對電子心理健康干預研究的映射審查側重於那些他們在方法中確定了以使用者為中心的設計的研究。然後使用英國設計委員會創新框架（https:// www.designcouncil.org.uk/our-work/news-opinion/double-diamond-universally-accepted-depiction-design-process/）中定義的步驟對論文進行分析，並從中確定了各種方法。只有 16 項研究提供了他們選擇的設計方法的定義，而 27 項研究中只有 5 項可以歸類為使用以使用者為中心的設計

^{30}

。

Discussion 討論

The consensus meeting identified a number of potential solutions to these challenges in the area of engagement. These cut across several different challenges and are organised below into broad themes and summarised in Table 2.
共識會議確定了在參與領域應對這些挑戰的許多潛在解決方案。這些挑戰涉及幾個不同的挑戰，並在下面分為廣泛的主題，並在表 2 中進行了總結。

The group did not aim to achieve a consensus on definitions of each term but agreed on the core concepts (Table 2, Theme 1a). Engagement is a complex term encompassing usage and adherence, but also cognitive, affective and motivational components. The group recognised the variety of different definitions and approaches which have already been proposed to describe engagement metrics, including conceptual frameworks for understanding in-the-moment engagement and how this could be used in designing strategies to promote engagement

^{7}

. Engagement can also be conceptualised and measured at the micro level (moment to moment usage and the user experience) and the macro level (including the depth of engagement with the behaviour change process)

^{31}

. These variations in definitions, as well as creating difficulties in comparisons across studies, may also explain at least some of the variability in the rates of engagement reported.
該小組的目標不是就每個術語的定義達成共識，而是就核心概念達成一致（表 2，主題 1a）。參與是一個複雜的術語，包括使用和依從性，但也包括認知、情感和激勵成分。該小組認識到已經提出的用於描述參與度指標的各種不同定義和方法，包括用於理解即時參與度的概念框架，以及如何將其用於設計促進參與度

^{7}

的策略。參與度也可以在微觀層面（即時使用和用戶體驗）和巨集觀層面（包括參與行為改變過程的深度）

^{31}

進行概念化和衡量。這些定義的差異，以及給研究之間的比較帶來困難，也可能至少部分解釋了報告的參與率的差異。

There was also consensus that agreed reporting standards are needed to allow comparison and synthesis of data across all individual research studies. The CONSORT-ehealth guidelines already include subitems relating to reporting attrition and engagement

^{32}

. However, reporting in individual studies still varies extensively

^{3}

and agreed guidance needs to be developed, standardised and implemented. An important part of these standards would be to report more than one engagement measure, including both objective (usage) and subjective metrics. Usage results should include raw data to allow direct comparisons, and other usage metrics (such as measures of adherence) should be reported transparently, including the pre-specified threshold used and justification for the rationale

^{11}

. Subjective measures of engagement allow convergent evidence to be compared to the behavioural data and should include not only commonly used questionnaires and self-report measures, but also newer metrics such as the Digital Working Alliance Inventory (D-WAI

^{33}

). This assesses the degree of alliance a user has to an app and has been shown to be associated with subjective and objective measures of app engagement and outcomes. Even with standardisation of metrics, however, individual study characteristics reporting will be essential: differing designs of
人們還一致認為，需要商定的報告標準，以允許對所有單獨研究的數據進行比較和綜合。CONSORT-ehealth 指南已經包括與報告流失和參與

^{32}

相關的子專案。然而，個別研究的報告仍然有很大

^{3}

差異，需要制定、標準化和實施商定的指南。這些標準的一個重要部分是報告多個參與度指標，包括客觀（使用方式）和主觀指標。使用結果應包括原始數據以允許直接比較，並且應透明地報告其他使用量度（如依從性度量），包括使用的預先指定的閾值和理由

^{11}

。參與度的主觀測量允許將收斂證據與行為數據進行比較，不僅應包括常用的問卷和自我報告措施，還應包括較新的指標，例如數位工作聯盟量表（D-WAI

^{33}

）。該指標評估了使用者對應用的聯盟程度，並已被證明與應用參與度和結果的主觀和客觀衡量有關。然而，即使指標標準化，單個研究特徵報告也是必不可少的：不同的設計

Table 2| Potential solutions to the challenges of engagement in digital mental health interventions identified by the consensus meeting
表 2|共識會議確定的參與數位心理健康干預挑戰的潛在解決方案

Theme 1 - definitions and terminology
主題 1 - 定義和術語
a) Standardisation of reporting of engagement in DHI research studies.
a）參與 DHI 研究報告的標準化。

The group did not achieve consensus on the exact definitions of each engagement term, but agreed on core concepts
該小組沒有就每個參與術語的確切定義達成共識，但就核心概念達成了一致
Engagement is a complex term encompassing usage and adherence, but also cognitive, affective and motivational components
參與是一個複雜的術語，包括使用和依從性，但也包括認知、情感和激勵成分
Agreed guidance needs to be developed, standardised and implemented by all studies
所有研究都需要制定、標準化和實施商定的指南
More than one engagement statistic should be reported (including both objective and subjective measures)
應報告多個參與度統計數據（包括客觀和主觀衡量）
Transparent reporting of raw data is needed to allow direct comparisons
需要透明地報告原始數據，以便進行直接比較
b) Assessment of the appropriate ‘dose’ of an intervention, to maximise engagement and outcomes.
b）評估干預的適當「劑量」，以最大限度地提高參與度和結果。
Short or ultra-short interventions may be appropriate in some cases
在某些情況下，短期或超短期干預可能是合適的
The target dose needs to be assessed for each intervention
需要評估每次干預的目標劑量

Theme 2 - demonstrating efficacy (outcomes) and cost effectiveness of effective engagement
主題 2 - 展示有效參與的功效（結果）和成本效益
a) Research studies need to be theory driven.
a）研究需要以理論為導向。
b) Research studies should actively report engagement and outcomes.
b）研究應積極報告參與度和結果。

Design trials to determine engagement, efficacy/outcomes and their facilitators
設計試驗以確定參與度、療效/結果及其促進因素
It is key to measure both patient-reported outcomes and experiences (PROMS and PREMS) as well as standard outcome measures
衡量患者報告的結果和體驗（PROMS 和 PREMS）以及標準結果測量是關鍵
More research is needed on the links between engagement and outcomes, including dose relationships and potential (bio)markers for optimal engagement and response.
需要對參與度和結果之間的聯繫進行更多研究，包括劑量關係和最佳參與度和反應的潛在（生物）標誌物。

Theme 3 - user involvement and user centred design
主題 3 - 用戶參與和以使用者為中心的設計
a) Improve standards of user involvement in DHI research studies, with more precise reporting.
a）提高使用者參與 DHI 研究的標準，提供更精確的報告。

Standardised guidance 標準化指南
Co-production and human-centred co-design
共同製作以人為本的共同設計
User involvement and engagement
用戶參與和參與
b) Investigate the mechanisms of engagement to identify the essential elements.
b）調查參與機制以確定基本要素。
Identifying the relative contribution of different engagement strategies
確定不同互動策略的相對貢獻
Maximising theory driven work in trustworthiness and engagement
在可信度和參與度方面最大化理論驅動的工作
c) Measure and report the potential harms of engagement.
c）衡量並報告參與的潛在危害。
d) Include clinicians and the wider workforce as users.
d）將臨床醫生和更廣泛的員工作為使用者。
interventions may mean that engagement can be measured only in certain ways in particular studies and the degree of personalisation may make defining the optimal dose more challenging.
干預措施可能意味著在特定研究中只能以某些方式來衡量參與度，而個人化程度可能使確定最佳劑量更具挑戰性。

Whilst sustained engagement may be needed for some interventions, longer engagement is not necessarily always better, and shorter interventions may be appropriate in some cases (e.g., single session interventions) (Table 2, Theme 1b). Researchers need to consider whether increased and sustained engagement is always required, by assessing the ‘target dose’. For some interventions, ultra-brief treatments may be possible. For example, in a study of an ultra-brief online treatment, self-reported anxiety and depression significantly reduced and the ultra-brief treatment was assessed as non-inferior to a standard-length treatment

^{34}

. Similarly, a guided singlesession online intervention was shown to be effective in reducing intrusive memories of work-related trauma

^{19, 20}

. Treatments could be shortened by only focussing on one aspect of a mental disorder (in this example by a focus on reducing intrusive memories in PTSD, rather than the whole symptomatology). This may not be suitable for all mental health conditions, but in some areas research could examine the potential to design shorter courses of digital treatments; rather than focussing on increasing usage it may be better in these cases to focus research efforts on decreasing the need for longer term usage and adherence.
雖然某些干預措施可能需要持續的參與，但較長時間的參與不一定總是更好，在某些情況下（例如，單次干預）可能適合較短的干預（表 2，主題 1b）。研究人員需要通過評估「目標劑量」來考慮是否總是需要增加和持續的參與。對於某些干預措施，超短期治療可能是可能的。例如，在一項超簡短在線治療的研究中，自我報告的焦慮和抑鬱顯著減少，並且超簡短治療被評估為不劣於標準長度治療

^{34}

。同樣，有指導的單次在線干預被證明可有效減少對工作相關創傷的侵入性記憶

^{19, 20}

。可以通過只關注精神障礙的一個方面來縮短治療時間（在這個例子中，重點是減少 PTSD 中的侵入性記憶，而不是整個癥狀）。這可能並不適合所有心理健康情況，但在某些領域，研究可以檢查設計更短數位治療療程的潛力;在這些情況下，與其專注於增加使用量，不如將研究工作集中在減少對長期使用和依從性的需求上。

Theory and rationale should direct both the research questions posed and the variables measured, guiding the direction and scope of enquiry (Table 2, Theme 2a). There was consensus that digital research, like its non-digital counterpart, should be guided by clear mechanistic or theoretical rationales. For example, studies should target specific psychological mechanisms with measurable outcomes, such as belief change, and these rationales should be explicitly reported to ensure transparency and scientific rigour

^{35}

. If the proposed theory is explicitly stated and the (bio) markers of interest are identified, this would enable comparison of individual studies. In addition, DHIs present an opportunity to design for more consistent targeting of causal mechanisms and the use of novel features to promote engagement with the mechanisms of change. For example, SloMo is a digitally supported psychological therapy for paranoia which supports visualisation of thinking habits (the key mechanism of change) as bubbles, which provide an engaging means of communicating subjective experiences while reducing information processing demands

^{27, 35, 36}

. Rather than simply adapting traditional therapy approaches into a digital format, mechanistic work is needed to identify the key processes in disorders in order to inform hypothesis-driven research in developing new and engaging DMHIs

^{37}

.
理論和基本原理應該指導提出的研究問題和測量的變數，指導調查的方向和範圍（表 2，主題 2a）。人們一致認為，數位研究與非數位研究一樣，應該以明確的機制或理論原理為指導。例如，研究應針對具有可衡量結果的特定心理機制，例如信念改變，並且應明確報告這些基本原理，以確保透明度和科學嚴

^{35}

謹性。如果明確陳述了所提出的理論並確定了感興趣的（生物）標誌物，這將能夠對單個研究進行比較。此外，DHI 提供了一個機會，可以設計更一致地針對因果機制，並使用新特徵來促進與變革機制的互動。例如，SloMo 是一種針對偏執狂的數位支援心理療法，它支援將思維習慣（變化的關鍵機制）可視化為氣泡，這提供了一種引人入勝的方式來交流主觀體驗，同時減少信息處理需求

^{27, 35, 36}

。與其簡單地將傳統治療方法改編成數位格式，不如需要機械工作來識別疾病的關鍵過程，以便為開發新的和引人入勝的 DMHI 的假設驅動研究提供資訊

^{37}

。

Trial design in digital health should be optimised to actively assess engagement, outcomes and their facilitators (Table 2, Theme 2b). Engagement with digital health tools and interventions is a complex behaviour which can be affected by internal facilitators (such as technology, gamification, design features) and external facilitators (such as using a digital navigator and a blended approach within a digital clinic environment)

^{38}

. Engagement can also be influenced by tailoring the intervention to a particular disease area or patient characteristics. For example, adapting an internet-based cognitive behavioural intervention to the needs of patients with depression in the context of multiple sclerosis (MS) is valued by patients and may increase usage and efficacy

^{39}

compared to a previous version that did not consider MS specific aspects. A recent study also investigated the impact of digital phenotyping to personalise app recommendation and suggested this resulted in increased engagement as measured by objective screen time and a measure of alliance, but more mechanistic studies are needed

^{24}

. Although common approaches to increase engagement include strategies such as personalisation and customisation, social and therapeutic support, within intervention guidance and real-time feedback

^{23}

, it is not clear which of these contribute most effectively to increased engagement and improved outcomes. Implementing some or all of these will increase costs and in-person time, thus potentially reducing some of the advantages of digital approaches over in-person consultation.
應優化數位健康的試驗設計，以積極評估參與度、結果及其促進因素（表 2，主題 2b）。參與數位健康工具和干預措施是一種複雜的行為，可能會受到內部促進者（例如技術、遊戲化、設計功能）和外部促進者（例如在數字診所環境中使用數位導航器和混合方法）

^{38}

的影響。根據特定疾病領域或患者特徵定製干預措施也可以影響參與度。例如，在多發性硬化症（MS）的情況下，根據抑鬱症患者的需求調整基於互聯網的認知行為干預受到患者的重視，並且與以前不考慮 MS 特定方面的版本相比，可能會增加使用和療效

^{39}

。最近的一項研究還調查了數位表型分析對個人化應用推薦的影響，並表明這導致了參與度的增加（通過客觀的屏幕時間和聯盟的衡量標準來衡量），但還需要

^{24}

更多的機制研究。儘管提高參與度的常見方法包括個人化和定製、社會和治療支援等策略，但在干預指導和實時反饋

^{23}

中，尚不清楚其中哪一項最有效地有助於提高參與度和改善結果。實施部分或全部這些將增加成本和面對面時間，從而可能會減少數位方法相對於面對面諮詢的一些優勢。

In terms of outcomes, it is key to measure both PROMS (patient reported outcome measures, for example symptoms, daily activity, quality of
就結果而言，衡量 PROMS（患者報告的結果測量，例如癥狀、日常活動、品質
life) and PREMS (patient reported experience measures such as satisfaction, communication/shared decision making, health literacy, autonomy and ease of access to healthcare) as well as standard objective measures of clinical outcomes, such as morbidity, mortality, or disease duration. Digital solutions are ideally suited to collect these and feed them back to the clinician and patient, for example, in cancer care

^{40}

or in mood disorders

^{41}

. Patients’ ‘empowerment’ (measures of the real-world outcomes of engagement) should also be reported across a variety of domains such as emotional and social wellbeing, self-management and control, education and knowledge, including health literacy and engagement in healthcare

^{42}

. Research should focus on the links between engagement, usage and outcomes, not only looking at ‘dose’ effects, but also the potential to identify markers for response from the early sessions. This will also rely on standardising reporting of engagement metrics (Theme 1).
life）和 PREMS （患者報告的經驗測量，如滿意度、溝通/共同決策、健康素養、自主性和獲得醫療保健的便利性）以及臨床結果的標準客觀測量，如發病率、死亡率或病程。數字解決方案非常適合收集這些資訊並將其反饋給臨床醫生和患者，例如，在癌症護理

^{40}

或情緒障礙

^{41}

中。患者的“賦權”（衡量參與的真實世界結果的衡量標準）也應在各種領域進行報告，例如情緒和社會福祉、自我管理和控制、教育和知識，包括健康素養和醫療保健

^{42}

參與。研究應側重於參與度、使用和結果之間的聯繫，不僅要關注“劑量”效應，還要關注從早期會議中確定反應標誌物的潛力。這也將取決於互動指標的標準化報告（主題 1）。

There was consensus that PPIE should be intentional and extend beyond a consultation model (Table 2, Theme 3a). The goal of PPIE is to ensure that DHIs meet the essential rights of users to be included in the development of relevant interventions and to ensure these are appropriately focussed on delivering real-world benefits for that particular population. Patients bring lived expertise, offering unique insights into their condition and the challenges of digital tools that may be overlooked by others. Representation from the full diversity of the target population aims to increase not just usage but also meaningful engagement and outcomes

^{25, 43}

and attempts to mitigate at least some of the impacts of the digital divide

^{44}

. Diversity includes adequate representation from underserved populations (e.g., older adults, individuals with lower digital literacy, from disadvantaged socioeconomic backgrounds or ethnic minorities) as well as those at the intersection of these categories. The following areas of focus were identified:
人們一致認為 PPIE 應該是有意識的，並超越諮詢模式（表 2，主題 3a）。PPIE 的目標是確保 DHI 滿足使用者的基本權利，以參與相關干預措施的制定，並確保這些干預措施適當地專注於為該特定人群提供現實世界的利益。患者帶來了生活專業知識，對他們的病情和數位工具可能被其他人忽視的挑戰提供了獨特的見解。來自目標人群的全部多樣性的代表不僅旨在增加使用，還旨在增加有意義的參與和結果

^{25, 43}

，並試圖減輕數字鴻溝的至少部分影響

^{44}

。多樣性包括來自服務不足人群（例如，老年人、數位素養較低的個人、來自弱勢社會經濟背景或少數族裔）以及處於這些類別交匯處的人群的充分代表性。確定了以下重點領域：

Standardised guidance: one barrier to greater user involvement may be that for digital approaches, including app design, there is, as yet, no standardised guidance on how to involve stakeholders, although frameworks have been proposed $^{25}$ . There are frameworks for user involvement more generally, such as the UK National Institute for Health and Care Research (NIHR) INVOLVE framework (https://www.nihr.ac.uk/ news/nihr-announces-new-standards-public-involvement-research), and the UK Design Council’s Double Diamond model (https://www. designcouncil.org.uk/our-resources/the-double-diamond/), but focused standardised guidance for designing and reporting user input in DHIs is needed.
標準化指南：提高用戶參與度的一個障礙可能是，對於包括應用程式設計在內的數位方法，儘管已經提出了 $^{25}$ 框架，但到目前為止還沒有關於如何讓利益相關者參與的標準化指南。更普遍地，有一些用戶參與的框架，例如英國國家健康與護理研究所（NIHR） INVOLVES 框架（https://www.nihr.ac.uk/ news/nihr-announces-new-standards-public-involvement-research）和英國設計委員會的雙鑽石模型（https://www.designcouncil.org.uk/our-resources/the-double-diamond/），但需要針對 DHI 中設計和報告使用者輸入的重點標準化指南。
Co-production and human-centred co-design: to ensure PPIE is meaningful, a co-production model is often used in clinical research. Although co-production indicates involvement, care needs to be taken as this approach typically focuses on developing and refining previously identified solutions to a previously agreed problem $^{26}$ . Whilst this can be helpful, an inclusive human centred co-design approach, including divergent and convergent phases in the development of DMHIs (https://www.designcouncil.org.uk/our-resources/ the-double-diamond/), may be needed to facilitate more effective engagement across diverse users. This method is distinct from coproduction in that it uses ethnographic methods to explore the needs and preferences of a diverse range of people, with iterative codesign of solutions to address these identified needs. The process aims to investigate 'what people need, rather than what they say they want $^{26}$ . This process aims to optimize the user experience and improve adherence for a diverse range of people $^{45}$ and involves a significant input of time, planning and funding, with only a few examples in mental healthcare so ${far}^{27}$ . Whilst user-centred design may improve engagement $^{28}$ , more direct evidence is needed. This might take the form of direct comparisons of different versions of DHIs or mechanistic studies to investigate key elements of engagement contributing to improved outcome. To assess how user centred codesign contributes to improved engagement and outcomes, studies need to report transparently the types of PPIE, who is involved and at what stage of product development with clearer definitions of the exact methodology used. Reporting of diversity among stakeholders is also a
聯合生產和以人為本的共同設計：為了確保 PPIE 的意義，臨床研究中經常使用聯合生產模式。雖然聯合生產表明參與，但需要小心，因為這種方法通常側重於開發和完善先前確定的解決方案，以解決先前商定的問題 $^{26}$ 。雖然這可能有所説明，但可能需要一種以人為本的包容性協同設計方法，包括 DMHI 開發中的發散和收斂階段（https://www.designcouncil.org.uk/our-resources/ 雙鑽/），以促進不同用戶之間更有效的參與。這種方法與聯合生產的不同之處在於，它使用人種學方法來探索不同人群的需求和偏好，並反覆運算協同設計解決方案來解決這些已確定的需求。該過程旨在調查“人們需要什麼，而不是他們說他們想要 $^{26}$ 什麼”。這個過程旨在優化用戶體驗並提高不同人群 $^{45}$ 的依從性，並涉及大量的時間、計劃和資金投入，只有少數幾個例子在心理健康方面 ${far}^{27}$ 。雖然以使用者為中心的設計可能會提高參與度 $^{28}$ ，但需要更直接的證據。這可能採取直接比較不同版本的 DHI 或機制研究的形式，以調查有助於改善結果的參與關鍵要素。為了評估以使用者為中心的協同設計如何有助於提高參與度和結果，研究需要透明地報告 PPIE 的類型、參與人員以及產品開發的哪個階段，並更清楚地定義所使用的確切方法。報告利益相關者之間的多樣性也是
challenge, and studies should include and report involvement of both expert PPIE and inclusive, diverse PPIE $^{25}$ . PPIE models should be carefully planned with sufficient allocated resources in terms of time and funding $^{25}$ . It is essential to investigate the relationship between different types of PPIE and outcomes, as the investment of time and resources could offset some of the potential advantages of digital approaches.
挑戰，研究應包括並報告專家 PPIE 和包容性、多樣化的 PPIE $^{25}$ 的參與。PPIE 模型應仔細規劃，並在時間和資金 $^{25}$ 方面分配足夠的資源。調查不同類型的 PPIE 與結果之間的關係至關重要，因為時間和資源的投入可能會抵消數位方法的一些潛在優勢。
User involvement and engagement: user involvement also needs to focus on meaningful engagement and outcomes (not just increased usage). This should include recognising the ‘engagement paradox’ and designing for disengagement once participants have reached their personal goals. Users may disengage for positive as well as negative reasons-these need to be tracked and reported in studies. Digital approaches have the potential to address some of the existing inequalities in care provision and to increase engagement in underserved populations, but to do this effectively they need to be proactively designed to mitigate issues exacerbating the digital divide $^{44, 45}$ .
用戶參與和參與：用戶參與還需要關注有意義的參與和結果（而不僅僅是增加使用量）。這應該包括認識到「參與悖論」並在參與者達到個人目標後進行脫離參與的設計。使用者可能會出於積極和消極的原因而退出 - 這些需要在研究中進行跟蹤和報告。數位方法有可能解決護理服務中的一些現有不平等問題，並增加服務不足人群的參與度，但要有效地做到這一點，需要積極設計它們以緩解加劇數位鴻溝的問題 $^{44, 45}$ 。

Mechanistic work is required, focussing specifically on the mechanisms underlying engagement and its translation to optimal outcomes, and whether there are identifiable (bio)markers for these (Table 2, Theme 3b). For example, some research suggests that working alliance and self-efficacy may be potential mediators between engagement and outcomes

^{46}

. Digital interventions offer advantages in terms of scalability, personalisation and integrated measurement of usage, but these can only be maximised if the elements which are essential for successful engagement are identified. Researchers need to consider what are the best approaches to study the mechanism of engagement in DMHIs, such as platform trials or Bayesian approaches and the use of analytic approaches such as machine learning and artificial intelligence

(AI)^{47}

. Frameworks such as the technology acceptance model (https://deepblue.lib.umich.edu/handle/2027.42/35547) can help identify key factors using broad categories of perceived usefulness, perceived ease of use, and actual use behaviour as relevant categories in determining engagement.
需要機械工作，特別關注參與及其轉化為最佳結果的機制，以及這些是否有可識別的（生物）標誌物（表 2，主題 3b）。例如，一些研究表明，工作聯盟和自我效能感可能是參與和結果

^{46}

之間的潛在仲介。數字干預在可擴充性、個人化和綜合使用方式測量方面具有優勢，但只有確定對成功參與至關重要的要素，才能最大限度地發揮這些優勢。研究人員需要考慮研究 DMHI 參與機制的最佳方法是什麼，例如平臺試驗或貝葉斯方法以及使用機器學習和人工智慧

(AI)^{47}

等分析方法。技術接受模型（https://deepblue.lib.umich.edu/handle/2027.42/35547）等框架可以幫助識別關鍵因素，使用感知有用性、感知易用性和實際使用行為等廣泛類別作為確定參與度的相關類別。

Mechanistic studies of engagement would allow the identification of the approaches which specifically improve engagement and outcomes. Strategies such as frequent contact, personalised feedback, gamification, and financial incentives can help reduce attrition rates

^{1, 48}

, and integrating these within digital interventions through automated notifications, prompts, and feedback has also shown promise

^{3, 49, 50}

. The digital space is also unique in that tools such as gamification, as well as increasing engagement, can also be a part of the therapy in themselves. For example, in gameChange, a VR therapy for agoraphobic avoidance in psychosis

^{51}

, users play a bubblepopping game in a virtual café. This allows users to test their fears about other people while building positive memories of social situations. Peer and therapist support (such as a 'peer digital navigator

^{52}

) for digital interventions can also promote better engagement. Support through social networks can increase engagement

^{9}

and may also have a positive effect on symptoms (for example, of depression)

^{41}

, although the exact mechanisms and essential ingredients are not fully elucidated. Personalisation of interventions can occur at several different levels to promote greater engagement and inclusivity: the DHI can be personalised and adapted to individual preferences and characteristics (within the technical specification), or more generally tailored according to their disease area, symptoms or age

^{53}

. For example, in DMHIs, mental health symptoms such as low motivation and impaired concentration as core features of anxiety and depressive disorders may directly affect engagement, as well as physical comorbidities such as fatigue, pain and sensory impairment

^{41}

. In addition, there can be personalisation in identifying which DHI is the correct fit for the individual (which may be facilitated by a staff member such as a ‘digital navigator’)

^{24}

. It is likely that a combination of different factors may be needed for a specific intervention or disease area. For example, in a scoping review of apps for schizophrenia, strategies used to improve engagement included push notifications and message prompts, personalisation, goal setting, gamification, multimedia formats, social connectedness, and support (peers and professionals)

^{5}

, but
參與度的機制研究將允許確定專門提高參與度和結果的方法。頻繁聯繫、個人化反饋、遊戲化和財務激勵等策略可以説明降低流失率

^{1, 48}

，通過自動通知、提示和反饋將這些整合到數位干預中也顯示出前景

^{3, 49, 50}

。數字空間的獨特之處還在於，遊戲化等工具以及增加參與度本身也可以成為治療的一部分。例如，在 gameChange 中，一種針對精神病中廣場恐懼症迴避的 VR 療法

^{51}

，使用者在虛擬咖啡館中玩泡泡爆破遊戲。這允許使用者測試他們對他人的恐懼，同時建立對社交場合的積極回憶。同伴和治療師對數位干預的支援（例如“同伴數位導航器

^{52}

”）也可以促進更好的參與。通過社交網路提供支援可以提高參與度

^{9}

，也可能對癥狀（例如，抑鬱症）

^{41}

產生積極影響，儘管確切的機制和基本成分尚未完全闡明。干預措施的個人化可以在幾個不同的層面進行，以促進更大的參與度和包容性：DHI 可以個人化並適應個人偏好和特徵（在技術規範內），或者更普遍地根據他們的疾病領域、癥狀或年齡

^{53}

進行定製。例如，在 DMHI 中，作為焦慮和抑鬱症核心特徵的心理健康癥狀（如動力不足和注意力不集中）可能直接影響參與度，以及疲勞、疼痛和感覺障礙等身體合併症

^{41}

。此外，在確定哪種 DHI 最適合個人時，可以進行個人化設置（這可能由“數字導航員”等工作人員提供便利）。

^{24}

特定干預或疾病領域可能需要不同因素的組合。例如，在對精神分裂症應用程式的範圍審查中，用於提高參與度的策略包括推送通知和消息提示、個人化、目標設定、遊戲化、多媒體格式、社會聯繫和支援（同行和專業人士），

^{5}

但
the individual contribution of each strategy and their relative contributions has not yet been assessed.
尚未評估每種策略的單獨貢獻及其相對貢獻。

It is also important to consider how to maximise the perceived ‘trustworthiness’ with DHIs, by harnessing the intrinsic benefits of what is already known about participant empathy and therapeutic relationships in the digital space. For example, there is a well-researched tendency for humans to anthropomorphise inanimate objects, which extends to digital interactions, including chatbots and conversational agents. Whereas online surveys show that a significant number of patients delay in person helpseeking because of embarrassment or a fear of being judged

^{54}

, rates of consultations with online platforms are high. For example, in a user survey of 2000 US adults,

67 %

of Americans said they had looked up their symptoms on an internet search engine and

52 %

had used a large language model like ChatGPT, looking for a diagnosis (https://www.usertesting.com/ resources/reports/consumer-perceptions-ai-healthcare). Transdisciplinary expertise in digital empathy is needed to maximise effective engagement - a degree of humanising the interface can improve engagement, but if ‘too human’ this may discourage disclosure of negative or embarrassing information

^{55, 56}

.
同樣重要的是，要考慮如何通過利用數字空間中已知的參與者同理心和治療關係的內在好處，最大限度地提高 DHI 的感知“可信度”。例如，人類有一種經過充分研究的趨勢，即將無生命的物體擬人化，這延伸到數位交互，包括聊天機器人和對話代理。雖然在線調查顯示，大量患者由於尷尬或害怕被評判而延遲親自尋求説明

^{54}

，但在線平台的諮詢率很高。例如，在對 2000 名美國成年人的用戶調查中，

67 %

美國人表示他們在互聯網搜尋引擎上查找了自己的癥狀，並使用

52 %

了像 ChatGPT 這樣的大型語言模型來尋找診斷（https://www.usertesting.com/ resources/reports/consumer-perceptions-ai-healthcare）。需要數位同理心方面的跨學科專業知識來最大限度地提高有效參與 - 一定程度的人性化介面可以提高參與度，但如果“太人性化”，這可能會阻止披露負面或令人尷尬的資訊

^{55, 56}

。

It is essential to measure and report potential adverse events actively in studies as they could also be a significant reason for dropout and loss of engagement

^{57}

(Table 2, Theme 3c). Adverse events could range from mild effects (such as frustration, boredom) to more severe (for example, symptom deterioration, suicidality or hospital admission). Their severity can also depend on the perceived impact on the patient; for example, screen time is often perceived as a potential adverse event, whereas the impact may be more nuanced and depend on the individual context (https://www. mqmentalhealth.org/mental-health-and-the-internet/).
在研究中積極衡量和報告潛在的不良事件至關重要，因為它們也可能是輟學和失去參與度

^{57}

的重要原因（表 2，主題 3c）。不良事件的範圍可以從輕微的影響（如沮喪、無聊）到更嚴重的（如癥狀惡化、自殺或住院）。它們的嚴重性也可能取決於對患者的感知影響;例如，屏幕時間通常被視為潛在的不良事件，而影響可能更加微妙，並且取決於個人情況（https://www。mqmentalhealth.org/mental-health-and-the-internet/）。

Just as participant user involvement should increase engagement, clinicians can also influence the adoption of digital interventions (Table 2, Theme 3d). Clinicians vary in their confidence and experience in the use of DMHIs. Increasing these would involve education and training in digital approaches

^{58}

and involving a diverse range of clinicians as ‘experts by experience’ at an early stage would facilitate better integration of DMHIs into the clinical pathway. Frameworks such as the Non-adoption, Abandonment, and challenges to Scale-up, Spread, and Sustainability (NASSS) framework

^{59}

can provide a structured approach for staff involvement, helping identify barriers and facilitators to real-world implementation throughout DMHI development and deployment. For example, the NASSS framework has been used to identify NHS staff views on the implementation of VR interventions on acute psychiatric wards, identifying both challenges (staff confidence with technology) and potential solutions (such as having a staff VR lead and accessible training)

^{60}

.
正如參與者的用戶參與應該增加參與度一樣，臨床醫生也可以影響數位干預的採用（表 2，主題 3d）。臨床醫生對 DMHIs 使用的信心和經驗各不相同。增加這些將涉及數位方法

^{58}

的教育和培訓，並在早期階段讓不同的臨床醫生成為“有經驗的專家”，這將有助於將 DMHI 更好地整合到臨床路徑中。不採用、放棄和擴大、傳播和可持續性挑戰（NASSS）框架

^{59}

等框架可以為員工參與提供一種結構化的方法，幫助識別整個 DMHI 開發和部署中實際實施的障礙和促進因素。例如，NASSS 框架已被用於確定 NHS 工作人員對在急性精神病房實施 VR 干預的看法，確定挑戰（員工對技術的信心）和潛在的解決方案（例如讓員工領導 VR 和無障礙培訓）。

^{60}

In this study we aimed to identify challenges and potential solutions in studying and enhancing digital health engagement, and how this might translate to improved outcomes for users and to focus the field for future research in this area. However, we are aware that there may be some potential limitations. While we conducted a systematic review, we limited our search to PubMed, which may have excluded relevant publications. We are also aware of the potential biases in the ways that industry and academic research sectors report and analyse engagement metrics, which may have affected the results reported

^{8}

. Additionally, like all consensus meetings, ours lacked standardized criteria for defining expertise. Although we selected participants to represent a diverse spectrum of views, the reliability of consensus opinions is dependent on the specialist knowledge and experiences of those who participated. We sought to ensure diverse perspectives by assembling an expert group and panel with varied expertise, nationalities, genders, ages, and disciplinary backgrounds. While the expert group included a wide range of experience in clinical research and real-world implementation in digital mental health, in future studies, we would also consider including contributions from commercial partners. In terms of lived experience, we included an expert with lived experience who made many material contributions throughout: to the literature review, presentation and discussion of the evidence, formation of consensus, and coproduction of the paper. In this way, we aimed to engage high-quality PPI
在這項研究中，我們旨在確定研究和增強數位健康參與的挑戰和潛在解決方案，以及這如何轉化為改善用戶的結果，並將該領域作為該領域未來研究的重點。但是，我們知道可能存在一些潛在的限制。雖然我們進行了系統綜述，但我們將檢索限制在 PubMed 上，這可能排除了相關出版物。我們還意識到，行業和學術研究部門在報告和分析參與度指標的方式上存在潛在偏差，這可能會影響報告

^{8}

的結果。此外，與所有共識會議一樣，我們的會議缺乏定義專業知識的標準化標準。儘管我們選擇的參與者代表了不同的觀點，但共識意見的可靠性取決於參與者的專業知識和經驗。我們試圖通過組建具有不同專業知識、國籍、性別、年齡和學科背景的專家組和小組來確保不同的觀點。雖然專家組在數位心理健康的臨床研究和實際實施方面擁有豐富的經驗，但在未來的研究中，我們還將考慮包括商業合作夥伴的貢獻。在生活經驗方面，我們包括了一位具有生活經驗的專家，他在整個過程中做出了許多實質性貢獻：文獻綜述、證據的呈現和討論、共識的形成以及論文的共同製作。通過這種方式，我們的目標是吸引高品質的 PPI
coproduction and we also identified a number of changes for future consensus meetings which could be implemented. For instance, in future meetings we will provide a glossary of terms with acronyms spelt out and lay definitions of scientific terminology to be used before and during the meeting to facilitate equal understanding. In addition, consistent with other examples of lived experience coproduction in mental health research

^{61}

, the digital topic will be chosen in collaboration with lived experience members.
聯合製作，我們還為未來的共識會議確定了一些可以實施的更改。例如，在未來的會議中，我們將提供一份術語表，其中拼寫出首字母縮略詞，並提供科學術語的通俗定義，以便在會議前和會議期間使用，以促進平等理解。此外，與心理健康研究中

^{61}

生活經驗合作的其他例子一致，數字主題將與生活體驗成員合作選擇。

This study utilised an international expert meeting, including lived experience and used a documented consensus method to incorporate the current state of evidence into our discussions. From this, we developed a consensus on the current challenges and next steps for assessing, recording and analysing engagement with DMHIs and their association with outcomes. Digital interventions have exciting potential in meeting the shortfall in service provision for participants with brain and mental health disorders. However, this can only be realised if we focus our efforts on high-quality standardised measures and reporting to identify which factors promote meaningful engagement and lead to more reliable realworld outcomes.
本研究利用了國際專家會議，包括生活經驗，並使用了記錄在案的共識方法將當前證據狀態納入我們的討論。由此，我們對評估、記錄和分析與 DMHI 的互動及其與結果的關聯的當前挑戰和後續步驟達成了共識。數字干預在填補為患有大腦和心理健康疾病的參與者提供服務的短缺方面具有令人興奮的潛力。然而，只有我們將精力集中在高質量的標準化措施和報告上，以確定哪些因素促進了有意義的參與並導致更可靠的現實世界結果，才能實現這一點。

Methods 方法

We used the consensus development panel (or consensus development conference (CDC)) approach and followed the methodology described and used by the US National Institutes of Health and the World Health Organization (www.who.int/publications/i/item/9789241548960)

^{62, 63}

. This is a particularly effective consensus method for identifying areas of challenge and potential solutions in a rapidly developing area and has been used in previous consensus studies in digital health

^{64, 65}

. Central to the methodology of the CDC is a face-to-face meeting between a group of individual experts and a separate panel of nonexpert participants, involving an interactive method to develop a consensus. The method enables a multidisciplinary approach, including lived experience, with all group members contributing to the discussion and recommendations and incorporating a literature review of the existing evidence.
我們使用了共識制定小組（或共識制定會議（CDC））方法，並遵循了美國國立衛生研究院和世界衛生組織（www.who.int/publications/i/item/9789241548960）

^{62, 63}

描述和使用的方法。這是一種特別有效的共識方法，用於確定快速發展領域的挑戰領域和潛在解決方案，並已用於數位健康的

^{64, 65}

先前共識研究。CDC 方法的核心是一組個人專家和一個單獨的非專家參與者小組之間的面對面會議，涉及一種互動方法來達成共識。該方法支援多學科方法，包括生活經驗，所有小組成員都參與討論和建議，並結合對現有證據的文獻綜述。

The consensus meeting 共識會議

The in-person meeting was held in Rome over 2 days in November 2024 and involved an international multidisciplinary group of individual experts (including with lived experience of mental health issues) and a separate panel of nonexpert participants (hereafter, “the panel”). In advance of the meeting, the panel conducted a systematic literature review using PubMed to search for papers relevant to the main themes identified by the experts (see Supplementary Note 1 for further details). This preliminary work formed the agenda for the questions to be addressed during the meeting. Consensus was defined as either fully met or unmet, with the outcome transparently reported

^{66}

. At the end of the meeting, the whole group engaged in plenary discussion to identify the key themes and structure the recommendations. The group identified challenges, which are outlined in the results, and potential solutions in the discussion.
面對面會議於 2024 年 11 月在羅馬舉行，為期 2 天，涉及一個由個人專家組成的國際多學科小組（包括具有心理健康問題生活經驗）和一個單獨的非專家參與者小組（以下簡稱“小組”）。在會議之前，該小組使用 PubMed 進行了系統的文獻綜述，以搜索與專家確定的主要主題相關的論文（有關詳細資訊，請參閱補充說明 1）。這項初步工作構成了會議期間要解決的問題的議程。共識被定義為完全滿足或未達成，結果透明地報告

^{66}

。會議結束時，整個小組進行了全體討論，以確定關鍵主題並構建建議。該小組確定了挑戰，這些挑戰在結果中進行了概述，並在討論中提供了潛在的解決方案。

The expert group 專家組

The 10 experts (AC, CB, SMG, EH, JK, S. Lambe, JT, PU, TW, AV) encompassed expertise in a variety of specialist areas within digital health (including virtual reality, coproduction and co-design, web-based screening and early intervention, digital approaches to empathy and the therapeutic relationship, philosophy, ethical issues, and lived experience). The group composition was gender-balanced, and professional backgrounds and experience included psychiatry, neurology, psychology, cognitive neuroscience, social sciences, methodology, evidence synthesis, regulatory pathways, patient and public involvement (PPI), philosophy and ethics. The expert group was international (including members from Germany, Luxembourg, Sweden, the United Kingdom, and the United States).
這 10 位專家（AC、CB、SMG、EH、JK、S. Lambe、JT、PU、TW、AV）涵蓋了數位健康各個專業領域的專業知識（包括虛擬實境、聯合生產和共同設計、基於網路的篩查和早期干預、同理心的數位方法和治療關係、哲學、倫理問題和生活體驗）。小組組成是性別平衡的，專業背景和經驗包括精神病學、神經學、心理學、認知神經科學、社會科學、方法、證據綜合、監管途徑、患者和公眾參與（PPI）、哲學和倫理學。該專家組是國際性的（包括來自德國、盧森堡、瑞典、英國和美國的成員）。

The panel 面板

The panel was composed of 10 members (KB, TG, CGS, KH, S. Lahutina, EGO, IP, SR, KAS, LT) and included early-career and more experienced
該小組由 10 名成員（KB、TG、CGS、KH、S. Lahutina、EGO、IP、SR、KAS、LT）組成，其中包括職業生涯早期和更有經驗的人
clinicians and researchers (at a different level of expertise). The panel members were chosen because they were well informed or experienced in mental health and digital interventions, but had no particular expertise in any one area of digital psychiatry. The panel was also international (including members from Germany, Italy, Luxembourg, Mexico, Sweden, Turkey, the United Kingdom, and the United States).
臨床醫生和研究人員（具有不同的專業水準）。之所以選擇小組成員，是因為他們在心理健康和數字干預方面有充分的瞭解或經驗豐富，但在數位精神病學的任何一個領域都沒有特別的專業知識。該小組也是國際性的（包括來自德國、義大利、盧森堡、墨西哥、瑞典、土耳其、英國和美國的成員）。

The panel increased their knowledge of the current evidence base in advance of the meeting by conducting a systematic literature review using PubMed to search for terms relevant to the main themes identified by the experts (see Supplementary Note 1 for further details). This preliminary work highlighted the areas of recent development, uncertainties, or challenges that formed the agenda for the questions to be addressed in the face-to-face meeting. Of the 261 papers identified by the panel in the systematic search, 11 (narrative and systematic reviews) were identified as essential reading (see Supplementary Fig. 1 for further details). All panellists were asked to read the selected papers and panel members were allocated to lead the group discussion on one expert talk to facilitate equal contributions from members of the expert group and the panel.
小組通過使用 PubMed 進行系統的文獻綜述來搜索與專家確定的主要主題相關的術語，從而在會議前增加了他們對當前證據基礎的瞭解（有關詳細資訊，請參閱補充說明 1）。這項初步工作突出了最近的發展、不確定性或挑戰領域，這些領域構成了面對面會議中要解決的問題的議程。在小組在系統檢索中確定的 261 篇論文中，有 11 篇（敘述性和系統評價）被確定為必讀讀物（更多詳細資訊見補充圖 1）。所有小組成員都被要求閱讀選定的論文，並分配小組成員在一次專家講座中主持小組討論，以促進專家組成員和小組成員的平等貢獻。

Reflexivity statement Reflexivity 語句

The meeting was convened by JT and AC, who selected the expert group to represent a balance of professional backgrounds, areas of specialist digital mental health expertise, lived experience, and gender. Panel members were suggested by members of the expert group and through professional contacts. The logistics of the meeting were supported externally by Angelini Pharma, but they did not have any input in the design of the meeting, identification or selection of the expert group or panel, agenda of the meeting, discussions, consensus, or output. Ethical approval was not required for this study as it did not involve research on human participants. The consensus was conference based, and all attendees offered contributions to the research topic in an open environment where talks were voluntary.
會議由 JT 和 AC 召集，他們選擇的專家組代表了專業背景、專業數位心理健康專業知識領域、生活經歷和性別的平衡。專家組成員由專家組成員和專業聯繫人推薦。會議的後勤工作得到了 Angelini Pharma 的外部支援，但他們在會議設計、專家組或小組的確定或選擇、會議議程、討論、共識或輸出方面沒有任何投入。這項研究不需要倫理批准，因為它不涉及對人類參與者的研究。共識是基於會議的，所有出席者都在一個開放的環境中為研究主題做出貢獻，演講是自願的。

Data availability 數據可用性

All data generated or analysed during this study are included in the published article and the Supplementary Information.
本研究期間生成或分析的所有數據都包含在已發表的文章和補充資訊中。

Received: 12 March 2025; Accepted: 6 June 2025;
收稿日期： 2025-03-12;錄用日期： 2025-6-6;

Published online: 02 July 2025
在線發佈： 2025 年 7 月 2 日

References 引用

Nwosu, A., Boardman, S., Husain, M. M. & Doraiswamy, P. M. Digital therapeutics for mental health: is attrition the Achilles heel?. Front. Psychiatry 13, 900615 (2022).
Nwosu， A.， Boardman， S.， Husain， M. M. & Doraiswamy， P. M. 心理健康的數位療法：損耗是阿喀琉斯之踵嗎？.前面。精神病學 13， 900615 （2022）。
Berardi, C. et al. Barriers and facilitators to the implementation of digital technologies in mental health systems: a qualitative systematic review to inform a policy framework. BMC Health Serv. Res. 24, 1-9 (2024).
Berardi， C. 等人。在心理健康系統中實施數位技術的障礙和促進因素：為政策框架提供資訊的定性系統評價。BMC Health Serv. Res. 24， 1-9 （2024）。
Elkes, J. et al. User engagement in clinical trials of digital mental health interventions: a systematic review. BMC Med. Res. Methodol. 24, 184 (2024).
Elkes， J. 等人。用戶參與數位心理健康干預的臨床試驗：系統評價。BMC Med. Res. Methodol.24, 184 (2024).
Torous, J., Michalak, E. E. & O’Brien, H. L. Digital health and engagement-looking behind the measures and methods. JAMA Netw. Open 3, e2010918 (2020).
Torous， J.， Michalak， E. E. & O'Brien， H. L. 數位健康和參與——尋找措施和方法背後的方法。美國醫學會網路。Open 3， e2010918 （2020）.
Simões de Almeida, R. & Marques, A. User engagement in mobile apps for people with schizophrenia: a scoping review. Front. Digit. Health 4, 1023592 (2023).
Simões de Almeida， R. & Marques， A. 精神分裂症患者移動應用程式的用戶參與度：範圍審查。前面。數位。健康 4， 1023592 （2023）。
Ng, M. M., Firth, J., Minen, M. & Torous, J. User engagement in mental health apps: a review of measurement, reporting, and validity. Psychiatr. Serv. 70, 538-544 (2019).
Ng， M. M.， Firth， J.， Minen， M. & Torous， J. 心理健康應用程式的使用者參與：測量、報告和有效性的審查。精神病學。第 70 期，538-544 （2019）。
Nahum-Shani, I., Shaw, S. D., Carpenter, S. M., Murphy, S. A. & Yoon, C. Engagement in digital interventions. Am. Psychol. 77, 836 (2022).
Nahum-Shani， I.， Shaw， S. D.， Carpenter， S. M.， Murphy， S. A. & Yoon， C. 參與數位干預。美國心理學 77， 836 （2022）。
Forbes, A., Keleher, M. R., Venditto, M. & DiBiasi, F. Assessing patient adherence to and engagement with digital interventions for depression in clinical trials: systematic literature review. J. Med. Internet Res. 25, e43727 (2023).
Forbes， A.， Keleher， M. R.， Venditto， M. & DiBiasi， F. 在臨床試驗中評估患者對數位干預的依從性和參與度：系統文獻綜述。J. Med. Internet Res. 25， e43727 （2023）。
Boucher, E. M. & Raiker, J. S. Engagement and retention in digital mental health interventions: a narrative review. BMC Digit. Health 2, 52 (2024).
Boucher， E. M. & Raiker， J. S. 數位心理健康干預中的參與和保留：敘述性回顧。BMC 數位。健康 2， 52 （2024）。
Lipschitz, J. M. et al. Digital mental health interventions for depression: scoping review of user engagement. J. Med. Internet Res. 24, e39204 (2022).
Lipschitz， JM 等人。抑鬱症的數位心理健康干預：用戶參與度的範圍審查。J. Med. Internet Res. 24， e39204 （2022）。
Sieverink, F., Kelders, S. M. & van Gemert-Pijnen, J. E. Clarifying the concept of adherence to eHealth technology: systematic review on when usage becomes adherence. J. Med. Internet Res. 19, e402 (2017).
Sieverink， F.， Kelders， S. M. & van Gemert-Pijnen， J. E. 澄清對電子健康技術的堅持概念：關於何時使用成為堅持的系統評價。J. Med. Internet Res. 19， e402 （2017）。
Milne-Ives, M., Homer, S., Andrade, J. & Meinert, E. The conceptualisation and measurement of engagement in digital health. Internet Interv. 36, 100735 (2024).
Milne-Ives， M.， Homer， S.， Andrade， J. & Meinert， E.數位健康參與的概念化和測量。互聯網互動 36, 100735 (2024).
Benbow, A. A. & Anderson, P. L. A meta-analytic examination of attrition in virtual reality exposure therapy for anxiety disorders. J. Anxiety Disord. 61, 18-26 (2019).
Benbow， A. A. & Anderson， P. L.對焦慮症的虛擬實境暴露療法中損耗的薈萃分析檢查。J. 焦慮症。61, 18-26 (2019).
Jardine, J. et al. Between rhetoric and reality: real-world barriers to uptake and early engagement in digital mental health interventions. ACM Trans. Comput.-Hum. Interact. 31, 1-59 (2024).
Jardine， J. 等人。在修辭與現實之間：數位心理健康干預的採用和早期參與的現實障礙。ACM Trans. Comput.-Hum.互動。31, 1-59 (2024).
Sanatkar, S. et al. Using cluster analysis to explore engagement and e-attainment as emergent behavior in electronic mental health. J. Med. Internet Res. 21, e14728 (2019).
Sanatkar， S. 等人。使用聚類分析來探索參與度和 e-achievement 作為電子心理健康中的緊急行為。J. Med. Internet Res. 21， e14728 （2019）。
Chien, I. et al. A machine learning approach to understanding patterns of engagement with internet-delivered mental health interventions. JAMA Netw. Open 3, e2010791 (2020).
Chien， I. 等人。一種機器學習方法，用於了解互聯網提供的心理健康干預的參與模式。美國醫學會網路。Open 3， e2010791 （2020）.
Milne-Ives, M., Homer, S. R., Andrade, J. & Meinert, E. Potential associations between behavior change techniques and engagement with mobile health apps: a systematic review. Front. Psychol. 14, 1227443 (2023).
Milne-Ives， M.， Homer， S. R.， Andrade， J. & Meinert， E. 行為改變技術與移動健康應用程式互動之間的潛在關聯：系統評價。前面。心理學 14， 1227443 （2023）。
Gan, D. Z., McGillivray, L., Han, J., Christensen, H. & Torok, M. Effect of engagement with digital interventions on mental health outcomes: a systematic review and meta-analysis. Front. Digit. Health 3, 764079 (2021).
Gan， D. Z.， McGillivray， L.， Han， J.， Christensen， H. & Torok， M. 參與數位干預對心理健康結果的影響：系統評價和薈萃分析。前面。數位。健康 3， 764079 （2021）。
Kanstrup, M. et al. A guided single session intervention to reduce intrusive memories of work-related trauma: a randomised controlled trial with healthcare workers in the COVID-19 pandemic in Sweden. BMC Med. 22, 403 (2024).
Kanstrup， M. 等人。減少對工作相關創傷的侵入性記憶的指導性單次干預：瑞典 COVID-19 大流行期間醫護人員的隨機對照試驗。BMC Med. 22， 403 （2024）。
Iyadurai, L. et al. Reducing intrusive memories after trauma via an imagery-competing task intervention in COVID-19 intensive care staff: a randomised controlled trial. Transl. Psychiatry 13, 290 (2023).
Iyadurai， L. 等人。通過 COVID-19 重症監護人員的圖像競爭任務干預減少創傷后的侵入性記憶：一項隨機對照試驗。精神病學 13， 290 （2023）。
Baumel, A., Edan, S. & Kane, J. M. Is there a trial bias impacting user engagement with unguided e-mental health interventions? A systematic comparison of published reports and real-world usage of the same programs. Transl. Behav. Med. 9, 1020-1033 (2019).
Baumel， A.， Edan， S. & Kane， J. M.是否存在影響使用者對無指導性電子心理健康干預的參與度的試驗偏倚？對已發佈的報告和相同程序的實際使用方式進行系統比較。翻譯 Behav.Med. 9， 1020-1033 （2019）。
Yardley, L. et al. Understanding and promoting effective engagement with digital behavior change interventions. Am. J. Prev. Med. 51, 833-842 (2016).
Yardley， L. 等人。瞭解並促進有效參與數位行為改變干預措施。美國醫學雜誌 51， 833-842 （2016）。
Balaskas, A., Schueller, S. M., Cox, A. L. & Doherty, G. Understanding users’ perspectives on mobile apps for anxiety management. Front. Digit. Health 4, 854263 (2022).
Balaskas， A.， Schueller， S. M.， Cox， A. L. & Doherty， G. 理解使用者對移動應用程式的焦慮管理的看法。前面。數位。健康 4， 854263 （2022）。
Dwyer, B. et al. Assessing digital phenotyping for app recommendations and sustained engagement: cohort study. JMIR Form. Res. 8, e62725 (2024).
Dwyer， B. 等人。評估應用推薦和持續參與的數位表型：佇列研究。JMIR 表格。第 8 號決議，e62725 （2024）。
Brotherdale, R., Berry, K., Branitsky, A. & Bucci, S. Co-producing digital mental health interventions: a systematic review. Digit. Health 10, 20552076241239172 (2024).
Brotherdale， R.， Berry， K.， Branitsky， A. & Bucci， S. 共同製作數位心理健康干預：系統評價。數位。健康 10， 20552076241239172 （2024）。
Vargas, C., Whelan, J., Brimblecombe, J. & Allendera, S. Co-creation, co-design and co-production for public health: a perspective on definitions and distinctions. Public Health Res. Pract. 32, e3222211 (2022).
Vargas， C.， Whelan， J.， Brimblecombe， J. & Allendera， S. 公共衛生的共同創造、共同設計和共同生產：定義和區別的視角。公共衛生研究實踐。32， e3222211 （2022）.
Hardy, A. et al. Co-designing technology to improve psychological therapy for psychosis: SloMo, a blended digital therapy for fear of harm from others. Schizophr. Res. 274, 526-534 (2024).
Hardy， A. 等人。共同設計技術以改善精神病的心理治療：SloMo，一種混合數位療法，用於擔心受到他人的傷害。精神分裂症。第 274 號決議，第 526-534 頁（2024 年）。
Kilfoy, A. et al. An umbrella review on how digital health intervention co-design is conducted and described. npj Digit. Med. 7, 374 (2024).
Kilfoy， A. 等人。關於如何進行和描述數位健康干預協同設計的總括性回顧。npj 數位。Med. 7， 374 （2024）。
Baines, R. et al. Meaningful patient and public involvement in digital health innovation, implementation and evaluation: a systematic review. Health Expect. 25, 1232-1245 (2022).
Baines， R. 等人。有意義的患者和公眾參與數位健康創新、實施和評估：系統評價。健康預期。25, 1232-1245 (2022).
Vial, S., Boudhraâ, S. & Dumont, M. Human-centered design approaches in digital mental health interventions: exploratory mapping review. JMIR Ment. Health 9, e35591 (2022).
Vial， S.， Boudhraâ， S. & Dumont， M. 數位心理健康干預中以人為本的設計方法：探索性映射審查。JMIR 門。健康 9， e35591 （2022）。
Short, C. E. et al. Measuring engagement in eHealth and mHealth behavior change interventions: viewpoint of methodologies. J. Med. Internet Res. 20, e292 (2018).
Short， CE 等人。衡量 eHealth 和 mHealth 行為改變干預的參與度：方法的觀點。J. Med. Internet Res. 20， e292 （2018）。
Eysenbach, G. & Consort-EHEALTH Group CONSORT-EHEALTH: improving and standardizing evaluation reports of web-based and mobile health interventions. J. Med. Internet Res. 13, e126 (2011).
Eysenbach， G. & Consort-EHEALTH Group CONSORT-EHEALTH：改進和標準化基於網路和移動的健康干預的評估報告。J. Med. Internet Res. 13， e126 （2011 年）。
Goldberg, S. B. et al. Alliance with an unguided smartphone app: validation of the digital working alliance inventory. Assessment 29, 1331-1345 (2022).
Goldberg， S. B. 等人。與無指導的智慧手機應用程式結盟：驗證數位工作聯盟庫存。評估 29， 1331-1345 （2022）。
Bisby, M. A., Balakumar, T., Scott, A. J., Titov, N. & Dear, B. F. An online therapist-guided ultra-brief treatment for depression and anxiety: a randomized controlled trial. Psychol. Med. 54, 902-913 (2024).
Bisby， M. A.， Balakumar， T.， Scott， A. J.， Titov， N. & Dear， B. F.在線治療師指導的抑鬱和焦慮超簡短治療：一項隨機對照試驗。心理學醫學 54， 902-913 （2024）。
Garety, P. et al. Digitally supported CBT to reduce paranoia and improve reasoning for people with schizophrenia-spectrum psychosis: the SlowMo RCT. Effic. Mech. Eval. 8, 1-90 (2021).
Garety， P. 等人。數字支援的 CBT 以減少偏執狂並改善精神分裂症譜系精神病患者的推理：SlowMo RCT。機械評估。8, 1-90 (2021).
Hardy, A. et al. How inclusive, user-centered design research can improve psychological therapies for psychosis: development of SlowMo. JMIR Ment. Health 5, e11222 (2018).
Hardy， A. 等人。包容性、以使用者為中心的設計研究如何改善精神病的心理療法：SlowMo 的開發。JMIR 門。健康 5， e11222 （2018）。
Holmes, E. A. et al. The lancet psychiatry commission on psychological treatments research in tomorrow’s science. Lancet Psychiatry 5, 237-286 (2018).
Holmes， E. A. 等人。柳葉刀精神病學委員會關於未來科學的心理治療研究。柳葉刀精神病學 5， 237-286 （2018）。
van Kessel, R., Ranganathan, S., Anderson, M., McMillan, B. & Mossialos, E. Exploring potential drivers of patient engagement with their health data through digital platforms: a scoping review. Int. J. Med. Inform. 5, 105513 (2024).
van Kessel， R.， Ranganathan， S.， Anderson， M.， McMillan， B. & Mossialos， E. 探索患者通過數位平台參與其健康數據的潛在驅動因素：範圍界定審查。Int. J. Med. Inform.5, 105513 (2024).
Gold, S. M. et al. Internet-delivered cognitive behavioural therapy programme to reduce depressive symptoms in patients with multiple sclerosis: a multicentre, randomised, controlled, phase 3 trial. Lancet Digit. Health 5, e668-e678 (2023).
Gold， SM 等人。互聯網提供的認知行為療法計劃以減輕多發性硬化症患者的抑鬱癥狀：一項多中心、隨機、對照、3 期試驗。柳葉刀手指。健康 5， e668-e678 （2023）。
Watson, L. et al. Utilizing patient reported outcome measures (PROMs) in ambulatory oncology in Alberta: digital reporting at the micro, meso and macro level. J. Patient-Rep. Outcomes 5, 1-8 (2021).
Watson， L. 等人。在艾伯塔省的門診腫瘤學中利用患者報告的結果測量（PROM）：微觀、中觀和巨集觀層面的數字報告。J. 患者代表。結果 5， 1-8 （2021）。
Borghouts, J. et al. Barriers to and facilitators of user engagement with digital mental health interventions: systematic review. J. Med. Internet Res. 23, e24387 (2021).
Borghouts， J. 等人。用戶參與數位心理健康干預的障礙和促進因素：系統評價。J. Med. Internet Res. 23， e24387 （2021）。
Fomo, M., Borga, L., Abel, T., Santangelo, PS., Riggare, S. & Klucken, J. Paccoud I Empowering Capabilities of People With Chronic Conditions Supported by Digital Health Technologies: Scoping Review. J Med Internet Res 27, e68458 (2025).
Fomo， M.， Borga， L.， Abel， T.， Santangelo， PS.， Riggare， S. & Klucken， J. Paccoud I 由數位健康技術支援的慢性病患者賦權能力：範圍界定審查。J Med 互聯網研究 27， e68458 （2025）。
Paccoud, I., Leist, A. K., Schwaninger, I., van Kessel, R. & Klucken, J. Socio-ethical challenges and opportunities for advancing diversity, equity, and inclusion in digital medicine. Digit. Health 10, 20552076241277705 (2024).
Paccoud， I.， Leist， A. K.， Schwaninger， I.， van Kessel， R. & Klucken， J. 推進數位醫學多樣性、公平性和包容性的社會倫理挑戰和機遇。數位。健康 10， 20552076241277705 （2024）。
Bond, R. R. et al. Digital transformation of mental health services. Npj Ment. Health Res. 2, 13 (2023).
Bond， RR 等人。心理健康服務的數字化轉型。NPJ 門。健康研究 2， 13 （2023）。
Hardy, A. et al. Bridging the digital divide in psychological therapies: observational study of engagement with the slowmo mobile app for paranoia in psychosis. JMIR Hum. Factors 9, e29725 (2022).
Hardy， A. 等人。彌合心理治療中的數位鴻溝：使用 slowmo 移動應用程式治療精神病偏執狂的觀察性研究。因素 9， e29725 （2022）。
Macrynikola, N., Chang, S. & Torous, J. Emotion regulation selfefficacy as a mechanism of alliance and outcomes in a brief, transdiagnostic digital mental health intervention: L’auto-efficacité de la régulation des émotions en tant que mécanisme d’alliance et de
Macrynikola， N.， Chang， S. & Torous， J. 情緒調節自我效能作為一種聯盟機制和結果在一項簡短的、跨診斷的數位心理健康干預中：情緒調節作為聯盟和自我效能機制的自我效能
résultats dans une brève intervention transdiagnostique numérique en santé mentale. Can. J. Psychiatry. 07067437241274201 (2024).
導致對心理健康的簡短數位跨診斷干預。Can. J. 精神病學。07067437241274201 (2024).
Joyce, D. W., Kormilitzin, A., Smith, K. A. & Cipriani, A. Explainable artificial intelligence for mental health through transparency and interpretability for understandability. npj Digit. Med. 6, 6 (2023).
喬伊斯，D. W.，科爾米利津，A.，史密斯，K. A.和奇普里亞尼，A. 通過透明度和可解釋性為可理解性，為心理健康提供可解釋的人工智慧。npj 數位。Med. 6， 6 （2023 年）。
Wu, A. et al. Smartphone apps for depression and anxiety: a systematic review and meta-analysis of techniques to increase engagement. npj Digit. Med. 4, 20 (2021).
Wu， A. 等人。抑鬱症和焦慮症的智慧手機應用程式：提高參與度的技術的系統評價和薈萃分析。npj 數位。Med. 4， 20 （2021）。
Gan, D. Z., McGillivray, L., Larsen, M. E., Christensen, H. & Torok, M. Technology-supported strategies for promoting user engagement with digital mental health interventions: a systematic review. Digit. Health 8, 20552076221098268 (2022).
Gan， D. Z.， McGillivray， L.， Larsen， M. E.， Christensen， H. & Torok， M. 促進使用者參與數位心理健康干預的技術支援策略：系統評價。數位。健康 8， 20552076221098268 （2022）。
Boucher, E. M., Ward, H. E., Mounts, A. C. & Parks, A. C. Engagement in digital mental health interventions: can monetary incentives help? Front. Psychol. 12, 746324 (2021).
Boucher， E. M.， Ward， H. E.， Mounts， A. C. & Parks， A. C. 參與數位心理健康干預：金錢激勵能説明嗎？前面。心理學 12， 746324 （2021）。
Freeman, D. et al. Automated virtual reality therapy to treat agoraphobic avoidance and distress in patients with psychosis (gameChange): a multicentre, parallel-group, single-blind, randomised, controlled trial in England with mediation and moderation analyses. Lancet Psychiatry 9, 375-388 (2022).
Freeman， D. 等人。自動虛擬實境療法治療精神病患者的廣場恐懼症迴避和痛苦（gameChange）：一項在英格蘭進行的多中心、平行組、單盲、隨機、對照試驗，包括中介和調節分析。柳葉刀精神病學 9， 375-388 （2022）。
Perret, S. et al. Standardising the role of a digital navigator in behavioural health: a systematic review. Lancet Digit. Health 5, e925-e932 (2023).
Perret， S. 等人。標準化數位導航器在行為健康中的作用：系統評價。柳葉刀手指。健康 5， e925-e932 （2023）。
Uhlhaas, P. & Torous, J. Digital tools for youth mental health. npj Digit. Med. 2, 104 (2019).
Uhlhaas， P. & Torous， J. 青少年心理健康的數位工具。npj 數位。Med. 2， 104 （2019）。
Levy, A. G. et al. Prevalence of and factors associated with patient nondisclosure of medically relevant information to clinicians. JAMA Netw. Open 1, e185293 (2018).
Levy， AG 等人。患者不向臨床醫生披露醫學相關信息的患病率和相關因素。美國醫學會網路。Open 1， e185293 （2018）。
Schuetzler, R. M., Giboney, J. S., Grimes, G. M. & Nunamaker, J. F. Jr. The influence of conversational agent embodiment and conversational relevance on socially desirable responding. Decis. Support Syst. 114, 94-102 (2018).
Schuetzler， R. M.， Giboney， J. S.， Grimes， G. M. & Nunamaker， J. F. Jr.對話代理體現和對話相關性對社會理想回應的影響。決定。支持系統 114， 94-102 （2018）。
Schuetzler, R. M., Grimes, G. M. & Giboney, J. S. The effect of conversational agent skill on user behavior during deception. Comput. Hum. Behav. 97, 250-259 (2019).
舒茨勒，R. M.，格萊姆斯，G. M.和吉博尼，J. S.欺騙期間對話代理技能對用戶行為的影響。計算。哼。行為。97, 250-259 (2019).
Linardon, J. et al. Systematic review and meta-analysis of adverse events in clinical trials of mental health apps. npj Digit. Med. 7, 363 (2024).
Linardon， J. 等人。心理健康應用程序臨床試驗中不良事件的系統評價和薈萃分析。Med. 7， 363 （2024）。
Smith, K., Torous, J. & Cipriani, A. Teaching telepsychiatry skills: building on the lessons of the COVID-19 pandemic to enhance mental health care in the future. JMIR Ment. Health 9, e37939 (2022).
Smith， K.， Torous， J. & Cipriani， A. 遠端精神病學技能教學：以 COVID-19 大流行的經驗為基礎，在未來增強心理健康護理。JMIR 門。健康 9， e37939 （2022）。
Greenhalgh, T. et al. Analysing the role of complexity in explaining the fortunes of technology programmes: empirical application of the NASSS framework. BMC Med. 16, 1-5 (2018).
Greenhalgh， T. 等人。分析複雜性在解釋技術計劃命運中的作用：NASSS 框架的實證應用。BMC Med. 16， 1-5 （2018）。
Brown, P. et al. Automated virtual reality cognitive therapy (gameChange) in inpatient psychiatric wards: qualitative study of staff and patient views using an implementation framework. JMIR Form. Res. 6, e34225 (2022).
Brown， P. 等人。住院精神病房的自動化虛擬實境認知療法（gameChange）：使用實施框架對工作人員和患者觀點進行定性研究。JMIR 表格。第 6 號決議，e34225 （2022）。
Smith, K. A. et al. Triangulating evidence from the GALENOS living systematic review on trace amine-associated receptor 1 (TAAR1) agonists in psychosis. Br. J. Psychiatry 20, 1-9 (2024).
Smith， KA 等人。來自 GALENOS 關於精神病中痕量胺相關受體 1 （TAAR1）激動劑的 GALENOS 活體系統評價的三角測量證據。Br. J. 精神病學 20， 1-9 （2024）。
Arakawa, N. & Bader, L. R. Consensus development methods: considerations for national and global frameworks and policy development. Res. Soc. Adm. Pharm. 18, 2222-2229 (2022).
Arakawa， N. & Bader， L. R. 共識發展方法：國家和全球框架和政策發展的考慮因素。Res. Soc. Adm. Pharm. 18， 2222-2229 （2022）。
Ferguson, J. H. The NIH consensus development program: the evolution of guidelines. Int. J. Technol. Assess. Health Care 12, 460-474 (1996).
弗格森，J. H.NIH 共識制定計劃：指南的演變。Int. J. Technol. 評估。醫療保健 12， 460-474 （1996）。
Smith, K. A. et al. Digital mental health: challenges and next steps. BMJ Ment. Health 26, e300670 (2023).
Smith， KA 等人。數位心理健康：挑戰和下一步。BMJ 門。健康 26， e300670 （2023）。
Smith, K. A. et al. Digital mental health for schizophrenia and other severe mental illnesses: an international consensus on current challenges and potential solutions. JMIR Ment. Health 11, e57155 (2024).
Smith， KA 等人。精神分裂症和其他嚴重精神疾病的數位心理健康：關於當前挑戰和潛在解決方案的國際共識。JMIR 門。健康 11， e57155 （2024）。
Gattrell, W. et al. ACCORD (ACcurate COnsensus Reporting Document): a reporting guideline for consensus methods in biomedicine developed via a modified Delphi. PLoS Med. 21, e1004326 (2024).
Gattrell， W. et al. ACCORD （ACcurate COnsensus Reporting Document）：通過修改後的 Delphi 開發的生物醫學共識方法報告指南。PLoS Med. 21， e1004326 （2024）。

Acknowledgements 確認

K.A.S., E.G.O. and A.C. are supported by the National Institute for Health and Care Research (NIHR) Oxford Health Clinical Research Facility. A.C. and E.G.O. are supported by an NIHR Research Professorship (grant RP-2017-08-ST2-006), the NIHR Oxford and Thames Valley Applied Research Collaboration, and the NIHR Oxford Health Biomedical Research Centre (grant BRC-1215-20005). A.C. is supported by Wellcome (Global Alliance for Living Evidence on Anxiety, Depression, and Psychosis [GALENOS] project). E.G.O. is supported by the Brasenose College Senior Hulme Scholarship. I.P. and J.K. are supported by the FNR Luxembourg (dHealthPD). E.A.H. receives support from the Swedish Research Council (2020-00873), the Oak Foundation (OCAY 18-442) and Wellcome Leap. TW acknowledges funding from the Maudsley Biomedical Research Centre at South London and Maudsley NHS Foundation Trust and King’s College London. T.G. is in receipt of a PhD studentship funded by the National Institute for Health and Care Research (NIHR) Maudsley Biomedical Research Centre (BRC). This paper is based on the discussion the authors had during a meeting which was held in Rome on November 18th-19th, 2024. The meeting was supported by Angelini Pharma. The sponsor did not have any influence on the content of the discussion, the outcome and the preparation of this manuscript.
K.A.S.、E.G.O. 和 AC 得到了美國國家健康與護理研究所（NIHR）牛津健康臨床研究設施的支援。AC 和 EGO 得到了 NIHR 研究教授職位（撥款 RP-2017-08-ST2-006）、NIHR 牛津和泰晤士河谷應用研究合作以及 NIHR 牛津健康生物醫學研究中心（撥款 BRC-1215-20005）的支援。A.C. 得到了 Wellcome（全球焦慮、抑鬱和精神病活證據聯盟 [GALENOS] 專案）的支援。E.G.O. 得到了 Brasenose College Senior Hulme 獎學金的支援。I.P. 和 J.K. 由 FNR 盧森堡（dHealthPD）提供支援。E.A.H. 得到了瑞典研究委員會（2020-00873）、橡樹基金會（OCAY 18-442）和 Wellcome Leap 的支援。TW 感謝南倫敦 Maudsley 生物醫學研究中心、Maudsley NHS 基金會信託基金和倫敦國王學院的資助。T.G. 正在接受由國家健康與護理研究所（NIHR）莫茲利生物醫學研究中心（BRC）資助的博士生獎學金。本文基於作者在 2024 年 11 月 18 日至 19 日在羅馬舉行的一次會議上的討論。本次會議得到了 Angelini Pharma 的支援。贊助商對討論的內容、結果和本稿件的準備沒有任何影響。

Author contributions 作者貢獻

Conceptualization was by A.C., J.T. and K.A.S. Literature search was completed by E.G.O., K.A.S., T.G., K.B., I.P., C.G.S., K.H., S.R., L.T. and S. Lahutina. The original manuscript was drafted by K.A.S., A.C. and J.T. All authors attended the consensus meeting and critically reviewed and approved the manuscript.
由 A.C.、J.T. 和 K.A.S. 進行概念化。文獻檢索由 E.G.O.、K.A.S.、T.G.、K.B.、I.P.、C.G.S.、K.H.、S.R.、L.T. 和 S. Lahutina 完成。原始手稿由 K.A.S.、A.C. 和 J.T. 起草。所有作者都參加了共識會議，並嚴格審查和批准了手稿。

Competing interests 利益爭奪

K.A.S., C.B., C.G.S., K.H., I.P., T.W., T.G., L.T., S.R. declare no competing interests. S.M.G. reports honoraria from Hexal, Angelini, and Tegus. PJU reports honoraria from Boehringer. E.A.H. reports honoraria from Angelini to attend the meeting at which this paper was discussed. E.A.H. receives occasional honoraria for keynotes and workshops, and royalties on 2 books. E.A.H. developed the ICTI intervention (ANEMONE™) and founded Afterimagery. AB. K.B. reports honoraria from Böhringer Ingelheim and from publishers and training institutes for workshops, books and lectures on psychotherapy. He is co-founder of two digital mental health start-ups. JK has shares in Portabiles HCT, Germany; reports advisory activities and honoraria from Angelini, Bial, Biogen, BMS, Celgene, Desitin, EverPharma, Lundbeck-Foundation, Medical Valley Digital Health Application Center, Novartis, RoxHealth, StreamedUp, Bauerfeind, Remepy. J.K. reports board activities for European national HTA bodies, Michael J Fox Foundation for Parkinson’s Research (MJFF), Bertelsmann Foundation, Germany, Hans Seidel Stiftung, Germany. AC has received research, educational and consultancy fees from INCiPiT (Italian Network for Paediatric Trials), CARIPLO Foundation, Lundbeck and Angelini Pharma. JT is the editor-in-chief of JMIR Mental Health and associate editor of npj Digital Medicine.
K.A.S.， C.B.， C.G.S.， K.H.， I.P.， T.W.， T.G.， L.T.， S.R. 聲明無競爭利益。S.M.G. 報告了 Hexal、Angelini 和 Tegus 的酬金。PJU 報告了 Boehringer 的酬金。E.A.H. 報告了 Angelini 參加討論本文的會議的酬金。E.A.H. 偶爾會收到主題演講和研討會的酬金，以及 2 本書的版稅。E.A.H. 開發了 ICTI 干預（ANEMONE™）並創立了 Afterimagery。AB. K.B. 報告了 Böhringer Ingelheim 以及出版商和培訓機構關於心理治療的研討會、書籍和講座的酬金。他是兩家數位心理健康初創公司的聯合創始人。JK 持有德國 Portabiles HCT 的股份;報告來自 Angelini、Bial、Biogen、BMS、Celgene、Desitin、EverPharma、Lundbeck-Foundation、Medical Valley Digital Health Application Center、Novartis、RoxHealth、StreamedUp、Bauerfeind、Remepy 的諮詢活動和酬金。J.K. 報告了歐洲國家 HTA 機構、Michael J Fox 帕金森病研究基金會（MJFF）、德國貝塔斯曼基金會、德國 Hans Seidel 基金會的董事會活動。AC 已收到 INCiPiT（義大利兒科試驗網路）、CARIPLO 基金會、靈北和 Angelini Pharma 的研究、教育和諮詢費用。JT 是 JMIR Mental Health 的主編和 npj Digital Medicine 的副主編。

Additional information 其他資訊：

Supplementary information The online version contains supplementary material available at
補充資訊在線版本包含補充材料，可從
https://doi.org/10.1038/s41746-025-01778-w.
https://doi.org/10.1038/s41746-025-01778-w。
Correspondence and requests for materials should be addressed to Katharine A. Smith.
信件和材料請求應寄給 Katharine A. Smith。

Reprints and permissions information is available at http://www.nature.com/reprints
重印本和許可權資訊可在 http://www.nature.com/reprints

Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
出版商注 Springer Nature 對已發佈地圖中的管轄權主張和機構隸屬關係保持中立。

Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article’s Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article’s Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creativecommons.org/licenses/by/4.0/.
開放獲取本文根據 Creative Commons Attribution 4.0 International License 獲得許可，該許可允許以任何媒體或格式使用、共用、改編、分發和複製，前提是您對原作者和來源給予適當的信任，提供指向 Creative Commons 許可的連結，並說明是否進行了更改。本文中的圖像或其他第三方材料包含在文章的知識共用許可中，除非在材料的致謝行中另有說明。如果材料未包含在文章的 Creative Commons 許可中，並且您的預期用途未被法律法規允許或超出允許的用途，您將需要直接從版權擁有者處獲得許可。要查看此許可證的副本，請訪問 http://creativecommons.org/licenses/by/4.0/。
© The Author(s) 2025
© 作者 2025

$^{1}$ Department of Psychiatry, University of Oxford, Oxford, UK. $^{2}$ Oxford Health NHS Foundation Trust, Warneford Hospital, Oxford, UK. $^{3}$ Oxford Precision Psychiatry Lab, NIHR Oxford Health Biomedical Research Centre, Oxford, UK. $^{4}$ Department of Psychology, Institute of Psychiatry, Psychology and Neuroscience, King’s College London, London, UK. $^{5}$ South London & Maudsley NHS Foundation trust, London, UK. $^{6}$ Department of Experimental Psychology, University of Oxford, Oxford, UK. $^{7}$ Department of Women’s and Children’s Health, Uppsala University, Uppsala, Uppsala County, Sweden. $^{8}$ Charité - Universitätsmedizin Berlin, Dept Psychiatry and Dept Psychosomatic Medicine, Campus Benjamin Franklin, Berlin, Germany. $^{9}$ German Center for Mental Health (DZPG), Partner Site Berlin/ Potsdam, Berlin, Germany. $^{10}$ INIMS, Universitätsklinikum Hamburg-Eppendorf, Hamburg, Germany. $^{11}$ Luxembourg Centre for Systems Biomedicine (LCSB), University of Luxembourg, Esch-sur-Alzette, Luxembourg. $^{12}$ Independent researcher (Patient and Public Involvement and Engagement (PPIE) representative), London, UK. $^{13}$ Centre Hospitalier de Luxembourg (CHL), Luxembourg, Luxembourg. $^{14}$ Dept. of Child and Adolescent Psychiatry, Psychosomatic Medicine and Psychotherapy, Charité - Universitätsmedizin, Berlin, Germany. $^{15}$ School of Psychology and Neuroscience, University of Glasgow, Glasgow, UK. $^{16}$ German National Academy of Sciences Leopoldina, Halle, Germany. $^{17}$ Centrum für Affektive Neurowissenschaften, Charité - Universitätsmedizin Berlin, Berlin, Germany. $^{18}$ Santa Giuliana Hospital, Verona, Italy. $^{19}$ Division of Digital Psychiatry, Beth Israel Deaconess Medical Center, Harvard Medical School, Boston, MA, USA. $^{20}$ These authors contributed equally: John Torous, Andrea Cipriani. $⊠$ e-mail: katharine.smith@psych.ox.ac.uk
$^{1}$ 牛津大學精神病學系，英國牛津。 $^{2}$ 牛津健康 NHS 基金會信託基金，沃內福德醫院，英國牛津。 $^{3}$ 牛津精準精神病學實驗室，NIHR 牛津健康生物醫學研究中心，英國牛津。 $^{4}$ 英國倫敦國王學院精神病學、心理學和神經科學研究所心理學系。 $^{5}$ 南倫敦和莫茲利 NHS 基金會信託，倫敦，英國。 $^{6}$ 牛津大學實驗心理學系，英國牛津。 $^{7}$ 瑞典烏普薩拉縣烏普薩拉烏普薩拉大學婦女和兒童健康系。 $^{8}$ 柏林夏里特醫學院 - 德國柏林本傑明佛蘭克林校區精神病學系和心身醫學系。 $^{9}$ 德國心理健康中心（DZPG），合作夥伴網站柏林/波茨坦，德國柏林。 $^{10}$ INIMS， Universitätsklinikum Hamburg-Eppendorf，漢堡，德國. $^{11}$ 盧森堡系統生物醫學中心（LCSB），盧森堡大學，盧森堡 Esch-sur-Alzette。 $^{12}$ 獨立研究人員（患者和公眾參與和參與（PPIE）代表），英國倫敦。 $^{13}$ 盧森堡醫院中心（CHL），盧森堡，盧森堡。 $^{14}$ 德國柏林夏里特醫學院兒童和青少年精神病學、心身醫學和心理治療系。 $^{15}$ 格拉斯哥大學心理學與神經科學學院，英國格拉斯哥。 $^{16}$ 德國國家科學院 Leopoldina，德國哈雷。 $^{17}$ Centrum für Affektive Neurowissenschaften， Charité - Universitätsmedizin Berlin， Berlin， Germany. $^{18}$ 義大利維羅納的聖朱利亞納醫院。 $^{19}$ 美國馬薩諸塞州波士頓哈佛醫學院貝絲以色列女執事醫療中心數位精神病學部。 $^{20}$ 這些作者的貢獻相同：John Torous、Andrea Cipriani。 $⊠$ 電子郵件： katharine.smith@psych.ox.ac.uk

Engagement and attrition in digital mental health: current challenges and potential solutions 數位心理健康的參與和流失：當前挑戰和潛在解決方案