The World Health Organization and other institutions are considering Artificial Intelligence (AI) as a technology that can potentially address some health system gaps, especially the reduction of global health inequalities in low- and middle-income countries (LMICs). However, because most AI-based health applications are developed and implemented in high-income countries, their use in LMICs contexts is recent and there is a lack of robust local evaluations to guide decision-making in low-resource settings. After discussing the potential benefits as well as the risks and challenges raised by AI-based health care, we propose five building blocks to guide the development and implementation of more responsible, sustainable, and inclusive AI health care technologies in LMICs.

Background: Developing a digital health innovation can require a substantial amount of financial and human resource investment before it can be scaled for implementation across geographical, cultural, and health care contexts. As such, there is an increased interest in leveraging eHealth innovations developed and tested in one country or jurisdiction and using these innovations in local settings. However, limited knowledge exists on the processes needed to appropriately adapt digital health innovations to optimize their transferability across geographical, cultural, and contextual settings. Objective: We report on the results of an adaptation study of Horyzons, a digital health innovation originally developed and tested in Australia. Horyzons is designed to prevent relapses and support recovery in young people receiving services for first-episode psychosis (FEP). The aim of this study is to assess the initial acceptability of Horyzons and adapt it in preparation for pilot testing in Canada. Methods: This research took place in 2 specialized early intervention clinics for FEP, located in 1 urban and 1 urban-rural setting, in 2 Canadian provinces. A total of 26 participants were recruited: 15 clinicians (age range 26-56 years) and 11 patients (age range 19-37 years). Following the digital health adaptation framework developed by our team, we used a mixed methods approach, combining descriptive quantitative and qualitative methods across 3 stages of data collection (focus groups, interviews, and consultations), analysis, and adaptations. Results: Overall, patients and clinicians appreciated the strengths-based approach and social media features of Horyzons. However, participants expressed concerns related to implementation, especially in relation to capacity (eg, site moderation, crisis management, internet speed in rural locations). They also provided suggestions for adapting content and features, for example, in relation to community resources, volume of text, universal accessibility (eg, for individuals with limitations in vision), and optimization of platform accessibility through mobile devices. Additional aspects of the innovation were flagged for adaptation during the final stages of preparing it for live implementation. These included terms of use, time zone configuration to reflect local time and date, safety and moderation protocols, the need help now feature, and the list of trigger words to flag posts indicative of potential risk. Conclusions: In the context of the COVID-19 pandemic and public health guidelines for social distancing, there is an increasing interest and need to leverage the internet and mobile technologies for delivering youth mental health services. As countries look to one another for guidance on how to navigate changing social dynamics, knowledge on how to utilize and adapt existing innovations across contexts is now more important than ever. Using a systematic approach, this study illustrates the methods, processes, results, and lessons learned on adapting a digital health innovation to enhance its local acceptability. International Registered Report Identifier (IRRID): RR2-10.2196/resprot.8810

This paper is a forum contribution that appeared in issue 51-2 in French. Due to the “positive buzz” it garnered following its publication, the MJE editorial team has made its translation available to our English readers. The original version can be accessed here: http://mje.mcgill.ca/article/view/9394 .

Digital health solutions that operate with or without artificial intelligence (D/AI) raise several responsibility challenges. Though many frameworks and tools have been developed, determining what principles should be translated into practice remains under debate. This scoping review aims to provide policymakers with a rigorous body of knowledge by asking: 1) what kinds of practice-oriented tools are available?; 2) on what principles do they predominantly rely?; and 3) what are their limitations? We searched six academic and three grey literature databases for practice-oriented tools, defined as frameworks and/or sets of principles with clear operational explanations, published in English or French from 2015 to 2021. Characteristics of the tools were qualitatively coded and variations across the dataset identified through descriptive statistics and a network analysis. A total of 56 tools met our inclusion criteria: 19 health-specific tools (33.9%) and 37 generic tools (66.1%). They adopt a normative (57.1%), reflective (35.7%), operational (3.6%), or mixed approach (3.6%) to guide developers (14.3%), managers (16.1%), end users (10.7%), policymakers (5.4%) or multiple groups (53.6%). The frequency of 40 principles varies greatly across tools (from 0% for 'environmental sustainability' to 83.8% for 'transparency'). While 50% or more of the generic tools promote up to 19 principles, 50% or more of the health-specific tools promote 10 principles, and 50% or more of all tools disregard 21 principles. In contrast to the scattered network of principles proposed by academia, the business sector emphasizes closely connected principles. Few tools rely on a formal methodology (17.9%). Despite a lack of consensus, there is a solid knowledge-basis for policymakers to anchor their role in such a dynamic field. Because several tools lack rigour and ignore key social, economic, and environmental issues, an integrated and methodologically sound approach to responsibility in D/AI solutions is warranted.

EHealth interventions have the potential to address challenges related to access, service engagement, and continuity of care in the delivery of mental health services. However, the initial development and evaluation of such interventions can require substantive amounts of financial and human resource investments to bring them to scale. Therefore, it may be warranted to pay greater attention to policy, services, and research with respect to eHealth platforms that have the potential to be adapted for use across settings. Yet, limited attention has been placed on the methods and processes for adapting eHealth interventions to improve their applicability across cultural, geographical, and contextual boundaries. In this paper, we describe an adaptation framework and protocol to adapt an eHealth intervention designed to promote recovery and prevent relapses in youth receiving specialized services for first-episode psychosis. The Web-based platform, called Horyzons, was initially developed and tested in Australia and is now being prepared for evaluation in Canada. Service users and service providers from 2 specialized early intervention programs for first-episode psychosis located in different provinces will explore a beta-version of the eHealth intervention through focus group discussions and extended personal explorations to identify the need for, and content of contextual and cultural adaptations. An iterative consultation process will then take place with service providers and users to develop and assess platform adaptations in preparation for a pilot study with a live version of the platform. Data collection was completed in August 2017, and analysis and adaptation are in process. The first results of the study will be submitted for publication in 2018 and will provide preliminary insights into the acceptability of the Web-based platform (eg, perceived use and perceived usefulness) from service provider and service user perspectives. The project will also provide knowledge about the adaptations and process needed to prepare the platform for evaluation in Canada. This study contributes to an important gap in the literature pertaining to the specific principles, methods, and steps involved in adapting eHealth interventions for implementation and evaluation across a diverse range of cultural, geographical, and health care settings.

While the conceptual foundations of Responsible Research and Innovation (RRI) were consolidated in the past decades, the practice of RRI remains poorly understood. The goal of our study was to gather the practical insights of professionals who design, develop and commercialize health innovations. We invited Canadian engineers, industrial designers, clinicians and entrepreneurs (n = 31) to visit a website describing nine innovations that possessed attributes of the Responsible Innovation in Health framework and to explain whether these examples illustrated important attributes of responsibility or not. Our qualitative analyses clarify how these innovators pondered whether: (1) stakeholder involvement is responsible; (2) businesses can behave responsibly; (3) innovations should adapt to health systems; and (4) the environment matters in health innovation. This study shows that innovators generally agree on the desirability of several responsibility principles, but identify feasibility issues that call for attention if RRI is to be meaningfully implemented in the health field.