PROTOCOL: Artificial Intelligence and Other Digital Tools Used by,and for Community Health Workers (CHWs) in Low and Middle-Income Countries (LMICs) to Improve Outcomes and Increase Effectiveness: An Evidence and Gap Map

Abstract

This is the protocol for the development of a Campbell Collaboration evidence and gap map (EGM). This protocol presents the methodology for developing the EGM. The objectives of this EGM include answering the following questions related to the use of Artificial Intelligence (AI) and other digital health tools by, and for CHWs in LMICs.

(1) What are the primary digital tools used by and for CHWs in LMICs, and for what purposes?

(2) To what extent, if any, is AI specifically being utilized by or for CHWs?

(3) What is the empirical evidence supporting the effectiveness of such tools, including AI, and the quality of that evidence?

(4) Does the use of digital tools and job aids by CHWs in LMICs contribute to (1) greater effectiveness and efficiency in carrying out assigned CHWs’ responsibilities; and (2) better health and other well-being outcomes for the clients and communities the CHWs serve?

(5) What are the risks and drawbacks of using digital tools, especially AI? What safeguarding strategies are employed to mitigate these risks?

(6) What are the primary gaps in the evidence for using digital tools, including AI, used by, and for CHWs in LMICs?

Keywords

artificial intelligence AI digital health community health workers CHW low and middle income countries,LMICs,mHealth,eHealth

Background

The Problem, Condition or Issue

As of early 2025, the World Health Organization (WHO) recognizes that community health workers (CHWs) are “well-placed to address the current 43 million health workers shortage.” (Community Health Impact Coalition, 2025). There are no accurate data quantifying how many CHWs exist, nor identifying how to locate and contact them. (Community Health Impact Coalition, 2025). CHWs generally live in rural areas, traveling to individual households within their communities, providing education, links to health services, and often diagnostics and health-related advice. Key challenges, pain points, and needs these increasingly important cadres of CHWs face in LMICs include.

(1) Identification and location of available CHWs, recruitment, and clear role definition;

(2) Accreditation, formal recognition, and professionalization of CHWs as part of national health systems, and as opportunities for advancement;

(3) Financing of CHWs (including how to pay CHWs) and incentives;

(4) Equipping CHWs (e.g., medical supplies, commodities, and personal protection equipment (PPE));

(5) Training (pre-service and in-service);

(6) Provision of structured supervision and access to feedback;

(7) Community involvement;

(8) Collection, management, and utilization of quality data; and

(9) Household visitation plans.

(Crigler et al., 2013; Matthias, 2022).

The cancellation in 2025 of much of the foreign funding supporting CHWs and the programs they implement or in which they are involved, has compounded obstacles for CHWs and the critical health access they extend to underserved communities. The withdrawal of foreign funding has led to job losses, reduced the number of essential health services that CHWs provide, and destabilized their personal and family finances. (Lotito, 2025).

These pressures highlight the importance and potential of leveraging digital and AI tools to reinforce programs delivered by CHWs and maintain service continuity.

The digital health landscape, particularly the rapidly expanding use of artificial intelligence (AI), should be regularly identified and assessed. Given the increasing reliance of national health systems upon CHWs (see, e.g., Ahmed et al., 2022), understanding how to best leverage the expanding number of digital tools is important for achieving maximum efficiency and health outcomes, as well as contributing to health equity within health systems of LMICs (Feroz et al., 2021a). Empowering CHWs who serve the most marginalized communities at the last mile, with the best tools, particularly when a large proportion of CHWs are found at the periphery of, or often not included in formal health systems, is a key strategy for achieving health equity and universal health coverage in LMICs (Community Health Impact Coalition, 2025; Last Mile Health, 2024).

The Intervention

We will be guided by the updated WHO Classification of Digital Interventions, Services and Applications in Health (Second Edition) (World Health Organization, 2023), which outlines the following user groups and data services: (1) persons; (2) health care providers; (3) health management and support personnel; and (4) data services.

Illustrative interventions using digital health, including AI, by and for CHWs may include workforce management, referral coordination, training, supervision, supply chain management, case management, decision support, disease surveillance, health communication, patient education, and data collection systems. These interventions may support functions related to service delivery, coordination of care, health system management, and communication between CHWs, supervisors, health facilities, and communities.

These interventions vary in duration, may be implemented by CHWs themselves, or on their behalf by supervisors, health authorities, health facilities, laboratories, including clinical service providers, community-based organizations, non-governmental entities, and community members.

Additional detail regarding the operationalization and classification of intervention types is provided below under Types of Interventions and Problems.

Why It Is Important to Develop the EGM

Globally, national public health systems are increasingly relying upon CHWs for the delivery of quality health services. This is particularly true in rural and other hard-to-reach areas where marginalized populations live, often referred to as the last mile. Many of these CHWs are not always formally recognized nor compensated under formal public health systems. A key focus for using digital health tools has been CHWs, one of the more important holders involved in the delivery of quality health services, particularly in LMICs. Notably, digital tools are being identified as a key strategy for addressing pain points and obstacles confronted by CHWs during care provision. While there have been past initiatives promoting the use of digital health for CHWs (e.g., “mPowering Frontline Health Workers”), evidence based on methodologically rigorous studies demonstrating the effectiveness of those digital health tools specifically targeting CHWs and their work, remains nascent (see, e.g., Whidden et al., 2018). Much of what has been documented regarding using digital health by or for CHWs has been either ad hoc, or public relations communications disseminated by organizations or companies vested in, or promoting a particular digital tool. There have been relatively few published analyses identifying and summarizing evidence supporting the use of digital health by and for CHWs (Braun et al., 2013), synthesizing health workers’ perceptions in using mHealth (Odendaal et al., 2020), focusing on the use of mobile digital health tools for particular tasks or under specific conditions (e.g., Amoakoh-Coleman et al., 2016; Borges do Nascimento et al., 2023; De Leeuw, 2019; Feroz et al., 2021a; Paduano et al., 2022; Palmer et al., 2020; World Economic Forum & Digital Health Action Alliance, 2023), or analyzing how community health worker programs can best be designed in general (World Health Organization, 2020). There are no recent comprehensive systematic literature reviews of methodologically rigorous academic studies taking stock and outlining quality evidence and documented outcomes, including, but not limited to, clinical health outcomes, connected specifically with using digital health by, and for CHWs.

As the use of digital health technologies rapidly expands and evolves, particularly as AI becomes a growing trend in digital health (Ciecierski-Holmes, T. et al., 2022; O’Donovan et al., 2022; Baxter et al., 2021; Li & Wu, 2025), identifying and assessing the digital health technologies landscape should occur regularly and frequently. The understanding among CHWs and their interest holders of the potential contributions and implications of AI remains relatively nascent (see, e.g., Igwama et al., 2024; Solano-Kamaiko et al., 2024; Islam et al., 2024). Integrating artificial intelligence (AI) inserts additional complexities in analyzing digital tools by and for CHWs. (See, e.g., Fischer, 2024; Duke Global Health Innovation Center, 2024; Igwama et al., 2024).

As LMICs’ national health systems increase their reliance on CHWs for health service delivery, understanding how to best leverage the growing number of digital tools is important for achieving maximum efficiency and improved health outcomes, as well as contributing to health equity within health systems. Furthermore, analyzing and understanding evidence related to AI, other digital tools, and CHWs can help promote recognition and avoidance of harms associated with digital technology in the “last mile”.

Empowering CHWs who serve the “last mile” and most marginalized communities with the evidence-based best tools, particularly when a large proportion of CHWs are found in the periphery of, or not included in formal health systems at all, arguably is a strategy in achieving health equity and universal health coverage in LMICs. A holistic analysis of digital health technologies used by and for CHWs necessitates a broad review of how digital health tools are implemented to achieve multiple health outcomes and to improve the overall job conditions of CHWs in LMICs.

Objectives

This EGM aims to answer the following questions related to the use of digital health tools by and for CHWs.

(1) What are the primary digital tools used by and for CHWs in LMICs, and for what purposes?

(2) To what extent, if any, is AI specifically being utilized by or for CHWs in LMICs?

(3) What is the empirical evidence supporting the effectiveness of such tools, including AI, and the quality of that evidence?

(4) Does the use of digital tools by CHWs in LMICs contribute to (a) greater effectiveness and efficiency in carrying out assigned CHWs’ responsibilities? (b) Improved health and other well-being outcomes among the clients and communities they serve?

(5) What are the risks and drawbacks of using digital tools, especially AI? What safeguarding strategies are employed to counter such risks?

(6) What are the primary gaps in the evidence for using digital tools, including AI, by and for CHWs in LMICs?

Methods

Evidence and Gap Map: Definition and Purpose

Digital health interventions, including those incorporating AI, are rapidly evolving and expanding in LMICs. The use of digital technologies often reflects global trends rather than being grounded in evidence derived from methodologically rigorous studies. The proliferation of digital interventions, particularly AI-enabled interventions used in conjunction with the work of CHWs, imposes on policymakers, funders, and implementers an obligation to discern which interventions are effective, under what circumstances, and consistent with the “do no harm” principle. This creates a pressing need for high-quality evidence to assess effectiveness, drawing on rigorously designed individual studies and systematic reviews. Recent bibliometric analyses indicate that the academic literature on AI in health has grown by approximately 20–30% per year over the past decade, with a 36-fold increase since 2000 (Shi et al., 2023; Xie et al., 2025). However, this growth in publications has not yet translated into a commensurate increase in methodologically rigorous impact evaluations, particularly within the contexts of LMICs.

As of this protocol’s development, we have identified very few methodologically rigorous studies and no comprehensive systematic reviews specifically assessing AI-enabled digital health interventions used by or for CHWs in LMICs. We did, however, find several publications addressing, describing, or promoting the use of AI by or for CHWs (see, e.g., O’Donovan et al., 2022; Baxter et al., 2021) and establishing guidelines for the use of AI multimodal models in health, including their potential applications for health workers (World Health Organization, 2024). A recent article in Lancet described in general how AI, as one example of innovations, can be used in prevention, diagnosis and treatment of HIV and other sexually transmitted infections. (Peters et al., 2025). Academic publications are publishing an increasing number of articles describing the potential of using AI in general for health. (see, e.g., Wankhede et al., 2025).

At least one literature review examined health care workers’ knowledge and perceptions of AI in Nigeria (Ogolodom et al., 2023). Existing reviews are narrow in scope and focus primarily on high-income regions (Anderson et al., 2019). Although some systematic reviews explore specific health outcomes (e.g., HIV, maternal health) (see, e.g., Ciecierski-Holmes et al., 2022), our formative scan revealed a dearth of rigorous studies and reviews examining digital tools used by or designed for CHWs.

Recent years have seen the emergence of the first registered or protocol-stage studies evaluating AI tools used by CHWs in LMICs. These include a prospective observational protocol in Rwanda assessing the use of large language models for CHW decision support (see, e.g., Menon et al., 2025) and a PATH-implemented study testing a generative-AI knowledge assistant for CHWs, as part of a broader AI-for-health initiative in Kenya, Nigeria, and Rwanda that includes workstreams involving CHWs and frontline health workers, including development of localized datasets from CHW clinical scenarios (PATH, 2025a). Although it is not focused on CHWs, a randomized controlled trial in Kenya is evaluating the use of an AI-assisted clinical decision support tool among primary care clinicians in Nairobi health facilities (PATH, 2025b).

While these ongoing trials mark significant progress, we have not yet found any completed randomized impact studies in which CHWs are the primary operators or beneficiaries of AI tools in LMICs. Related facility-based evidence, such as the PROSPECT trial in Malawi using AI computer-aided detection (CAD) for tuberculosis triage, demonstrates potential health system benefits but does not directly evaluate CHW-led interventions (MacPherson et al., 2021; see also Rodrigues et al., 2022).

Given this context and the broad scope of digital health and CHW interventions, an EGM is a critical first step in identifying clusters of evidence that may later be synthesized in systematic reviews. It is essential to determine the nature and extent of the existing evidence and assess whether meta-analysis is feasible. For this reason, a traditional systematic review would be too limited in scope to capture the full evidence landscape of digital health and CHWs in LMICs. We anticipate that few methodologically rigorous studies exist, reinforcing the need for an EGM to systematically identify and highlight existing evidence and gaps (see, e.g., White et al., 2020; Philbrick et al., 2022). This work will support key interest holders and advocates in setting research priorities and identifying evidence clusters for future systematic reviews.

The evidence base on the use, effectiveness, and feasibility of digital health in the roles of CHWs in LMICs remains nascent. Compared with a systematic review, an EGM is a tool for understanding this evidence landscape, describing where digital health work (by and for CHWs) is being conducted (context); identifying focus populations, implementation strategies, and the quality and diversity of existing evidence. An EGM guides future research priorities and informs resource allocation and investment decisions. Finally, an EGM that maps evidence related to the operationalization and implementation of digital health in the context of CHWs within LMICs is essential for implementers to uphold the do no harm principle when delivering interventions in particularly vulnerable settings.

Framework Development and Scope

The final framework will be developed through a systematic landscape review examining the various ways in which digital health tools are used by and for community health workers (CHWs), with a primary focus on low- and middle-income countries (LMICs). For comparative purposes, we may also consider examples from higher-income settings, which will be discussed in the Discussion section of the final EGM. To complement the literature review, we will conduct key informant consultations with implementation interest holders such as international and national NGOs who work directly with CHWs. These discussions will help to clarify CHWs’ roles, responsibilities, and operational challenges. Several members of the research team are affiliated with organizations that collaborate closely with CHWs, positioning us to contribute meaningfully to defining the framework’s scope and parameters. The framework will also draw on recent mapping exercises that document the use of digital health tools by and for CHWs across both LMICs and higher-income contexts (e.g., United Nations Development Programme). In addition, a recent consultancy commissioned by CARE USA, which examined the potential contribution of digital health to CHWs’ functions provides further evidence to inform the framework (Matthias, 2022). Finally, our engagement with the Community Health Impact Coalition (CHIC), an alliance promoting research and best practices in CHW programs, provides additional insight into ongoing CHW-related research and implementation efforts in LMICs.

Interest Holder Engagement

We are engaging with various interest holders in connection with the preparation of this protocol, including implementing organizations that currently incorporate CHWs within their programs and have an interest in identifying quality evidence related to digital health. CHIC has been one such important interest holder with which we have consulted for research guidance thus far during the development of the EGM protocol.

We will consult with strategists and program managers within the digital health industry who are working with digital health technologies including AI (e.g., NVIDIA, Anthropic, Vercel, AWS). We will also engage with organizations such as Dimagi, Medic, NetHope, international organizations such as UNICEF, and local implementing organizations. We also intend to consult with experts and advisors at the World Health Organization’s (WHO) Health Workforce Unit on an ad hoc basis. These resources, along with our direct professional experience working with CHWs in LMIC contexts, have been beneficial in informing and shaping the EGM. Currently, there is no formal Advisory Group, but the possibility of forming and engaging one remains if funding can be secured.

Conceptual Framework

The final dimensions of the EGM will be primarily informed by WHO classifications, the on-the-ground practical experience of organizations using digital health, and research identifying the types of interventions that augment the roles of CHWs and address the gaps they face (see, e.g., Living Goods et al., 2019; Matthias, 2022). Experience during the COVID-19 response has also provided a foundation for identifying relevant dimensions (see, e.g., Aranda et al., 2024; Baxter et al., 2021; Feroz et al., 2021a, 2021b; Helldén et al., 2023).

Below is a sample illustration of the Theory of Change for the use of digital health by and for CHWs.

Dimensions

We are developing a coding sheet to systematically capture key dimensions of information extracted from studies included in the EGM. These dimensions are designed to support structured analysis of digital health interventions, including those incorporating artificial intelligence, and are informed by published literature, implementation reports, and our practical experience.

The following dimensions of information to be extracted from included studies include.

(i) Type of organization producing the study report (e.g., academic institution, government entity, contract research firm);

(ii) Target populations of the digital health interventions (e.g., CHWs, broader health workforce, health system, patients);

(iii) Primary users of digital health (e.g., CHWs, supervisors, health facility staff, health authorities, clients);

(iv) Geographic setting (region and country where study was conducted);

(v) Population setting (urban, rural, peri-urban, mixed, unknown);

(vi) Sample size;

(vii) Type of digital health platform or modality (e.g., mobile phone, tablet, web-based laptop or PC, AI-enabled system, other);

(viii) Health system challenge(s) addressed (e.g., access, quality of care, workforce capacity, supply chain, surveillance); - WHO-aligned intervention, operationalized in this EGM through the following coding categories: 1 = targeted communication to persons (e.g., communicating test results); 2 = untargeted communications to persons; 3 = person-to-person communication; 4 = health tracking; 5 = health reporting; 6 = consent management; 7 = registration; 8 = decision support; 9 = health-care provider communication/consultation; 10 = referral coordination; 11 = scheduling or planning; 12 = training; 13 = prescription and medicine management; 14 = financial transactions; 15 = human resource management; 16 = inventory and supply management; 17 = public health event notification; 18 = CRVS (civil registration and vital statistics); 19 = certification; 20 = health worker supervision; and 21 = data collection or management (World Health Organization, 2023);

(ix) Gender (e.g., whether gender of CHWs or clients is analyzed or reported);

(x) Unintended or negative consequences;

(xi) Study duration;

(xii) Methods used;

(xiii) Output/outcome measured; and

(xiv) Summary of findings.

Types of studies to be identified and considered for this EGM include:

(1) Systematic reviews of experimental or quasi-experimental studies;

(2) Experimental studies (e.g., randomized controlled trials and other experiments with random assignment), including quasi-randomized controlled trials (QRCTs), where participants are allocated using methods such as alphabetical allocation without revealing personally identifiable information; non-randomized controlled trials (NRCTs), where participants are allocated through actions controlled by the researcher; and non-randomized studies, where allocation is not controlled by the researcher and two or more groups of participants are compared;

(3) Quasi-experimental studies with well-defined comparison groups, including non-randomized controlled trials, cohort studies, case-control studies, and cross-sectional analytical studies; and

(4) Rigorously designed qualitative research guided by precise and clear research questions (RQs), with adequate data collection to address the RQs, interpretation of results substantiated by data, and coherence among qualitative data sources, collection methods, analysis, and interpretation. Descriptive qualitative evidence related to implementation or cost-effectiveness of digital platforms is particularly relevant.

We may include studies without comparison groups, including case studies, if they fulfill at least one of the following criteria.

(1) Examination of elements of digital health implementation that are relevant to CHWs and related interest holders;

(2) Examination of cost-effectiveness; or

(3) Evaluation using a critical appraisal tool that has been tested for internal consistency, reliability, and validity, such as the Quality Assessment Tool for Quantitative Studies (see, e.g., Fitzpatrick-Lewis et al., 2009, citing Greenhalgh et al., 2005; Whittemore & Knafl, 2005; Norris & Atkins, 2005; Thomas et al., 2024; Zaza et al., 2000).

Such studies will be clearly identified in the EGM.

Types of Interventions and Problems

The EGM will be guided by the updated WHO Classification of Digital Interventions, Services and Applications in Health (Second Edition), which outlines the following user groups and data services: (1) persons, (2) health care providers, (3) health management and support personnel, and (4) data services (World Health Organization, 2023).

The World Health Organization defines digital health as “digital technologies used to address health needs, which are a fundamental part of health priorities that should be ethical, safe, and equitable… The term encompasses a wide range of applications, from telemedicine and mobile health apps to big data and AI, all used to improve health services and outcomes” (World Health Organization, 2019).

Digital health is rooted in eHealth, which the WHO defines as “the use of information and communications technology (ICT) in support of health” (World Health Organization, 2019). Key technologies comprising digital health include “mobile applications (mHealth) for tracking health, telemedicine for remote consultations, wearable devices, and the use of big data, genomics, and AI” (World Health Organization, 2019).

Artificial intelligence (AI) is “the use of computers and technology to simulate intelligent behavior and critical thinking comparable to that of a human being” (Amisha et al., 2019). The intervention categories are aligned with the World Health Organization’s Classification of Digital Health Interventions (2^nd ed.) (World Health Organization, 2023).

Illustrative interventions using digital health, including AI, by and for CHWs include:

• Recruitment and performance monitoring of CHWs (human resources for health [HRH] management systems);

• Identification and geolocation of CHWs (health workforce registries and GIS-enabled systems);

• Registration of CHWs (electronic health workforce information systems);

• Referral coordination and tracking (electronic referral systems);

• Pre-service, in-service, and refresher training (digital training and eLearning systems);

• Administrative and care coordination tasks (digital workflow and task management systems)

• Supportive supervision (digital supervision and performance management tools;

• Remuneration and incentive management (digital financial services and payment systems);

• Inventory and supply chain management (logistics management information systems [LMIS]);

• Case management (electronic case management systems, including patient registration, service tracking, referrals, follow-up, and GIS-enabled tracking);

• Household visitation planning and follow-up (digital scheduling and routing systems);

• Case closure documentation (EMR or case reporting systems);

• Diagnostic and clinical decision support (digital decision-support systems and clinical algorithms);

• Health data collection and reporting (digital data collection and health information systems);

• Patient education and behavior change communication (digital health education, IVR, and messaging platforms);

• Communication of clinical results (digital laboratory and results reporting systems);

• Disease and treatment management (digital adherence and remote monitoring systems);

• Outbreak and disease surveillance (digital surveillance and reporting systems);

• Assessments and screenings (digital screening and assessment tools); and

• Performance feedback and quality assurance (digital dashboards, scorecards, and audit systems).

When considering whether digital health promotes CHWs’ “efficiency”, which can also be synonymous with “productivity”, we will identify studies that track metrics such as: patient volume, task completion rates, medication administration accuracy, adherence to clinical guidelines, patient satisfaction scores, wait times, and staff-to- patient ratios, while also considering factors like resource utilization, workload management, and quality of care delivered. (Hasan et al., 2021; Zarska et al., 2021).

These interventions last for variable durations and are conducted/delivered by CHWs themselves, on their behalf, or on behalf of their supervisors and health authorities, health facilities and laboratories, including clinical service providers, and community-based organizations, non-governmental entities, as well as individual community members. The EGM will examine: (1) interventions in which CHWs use digital tools themselves and (2) CHWs’ interventions in which digital tools are used directly by any other interest holder in the intervention (supervisors, program coordinators, patients, etc.) in relation to CHWs’ work and responsibilities.

Types of Population

Participants (or populations) in studies that will be considered include CHWs, primarily women, working in LMICs’ settings. We will refer to the definition used in a recent systematic review published in The Lancet Global Health: CHWs were defined as lay health workers who: (1) are primarily based in the community (as opposed to a primary health facility); (2) perform tasks related to health-care delivery, and (3) have received organized training but have no formal or paraprofessional certification of tertiary education degree (i.e. this includes traditional birth attendants [TBAs] trained for an intervention but untrained TBAs, nurses, and midwives are excluded)”.

Community health workers (CHWs) are health care providers who live in the communities they serve and receive lower levels of formal education and training than professional health care workers such as nurses and doctors. (World Health Organization, 2020). Different names used for CHWs in different countries and contexts include Accredited Social Health Activists, Community Health Assistants, Community Health Extension Workers, Community Health Promoters, Community Health Volunteers, Health Educators, Health Extension Workers, Lay Health Workers, Lady Health Workers, Promotoras de Salud, Village Health Team members, and Ward based Outreach Team members. (see, e.g., Living Goods et al., 2019).

We will also, for purposes of the EGM, consider CHWs as those who deliver care services that address most of a person’s health needs throughout their lifetimes, including physical, mental, and social well-being. Foundational to primary health care are integrated people-centered health services that are managed and delivered across a continuum of services (health promotion, disease prevention, diagnosis, treatment, disease-management, rehabilitation, and palliative care), delivered within and beyond the health sector, throughout the life course (World Health Organization, 2006).

Care, according to the WHO, includes a broad range of personal, social, and health services and support for to individuals with limited or declining functional ability due to aging, mental or physical illness, or disability.

We will align our definitions with those used in the WHO’s World Health Report 2006. The report defines health workers broadly as inclusive of family caregivers, patient-provider partners, part-time workers (especially women), health volunteers, and community workers (World Health Organization, 2006). Excluded from searches will be frontline health workers (FLHWs) who work full-time exclusively in primary health facilities.

Types of Outcome Measures

Outcome measures will be derived from logic models related to the use/application of digital tools and platforms identified under Types of Interventions and Problems. Primary outcomes (evidence of: (1) greater effectiveness and efficiency in carrying out assigned CHWs’ responsibilities, (2) better health outcomes for the clients and communities the CHWs serve): Primary and secondary outcomes for the EGM include, but are not limited to.

(1) CHWs’ knowledge of health-related issues and other changes in capacity;

(2) CHWs’ population and disease coverage;

(3) Numbers, accuracy, and completion of referrals;

(4) Clinical outcomes related to disease and health ailments (indicators of mortality and morbidity in communities which that are served by CHWs) in maternal, newborn, and child health, sexual and reproductive health (e.g., family planning services, prevention of, and access to sexual and gender-based violence response services); HIV and AIDS (e.g., treatment adherence, prevention of mother-to-child transmission (PMTCT) outcomes, infectious diseases, prenatal, perinatal, and postnatal care);

(5) Access to quality health services (including facility-based births);

(6) Accuracy, completeness, and timeliness of data collection;

(7) Supply chain and logistics (frequency of stock-outs);

(8) Management and monitoring of cases, including tracking of services;

(9) Frequency of hospitalizations;

(10) Frequency and quality of home visitations;

(11) CHWs workload, productivity, and time efficiency;

(12) Quality of care delivered by CHWs (including adherence to clinical guidelines); and

(13) Usability and performance of digital systems.

Secondary outcomes (or intermediate output and outcomes that also can contribute to, or are correlated with Primary Outcomes):

(14) Patient satisfaction;

(15) Other interest holder perceptions of CHW performance and health services;

(16) CHWs’ attitudes and perceptions;

(17) Geographic information system (GIS) mapping of communities/households;

(18) Policy and process changes;

(19) Changes in social or cultural norms, attitudes, and/or behaviors;

(20) Changes in levels of digital literacy; and

(21) Interoperability of digital systems.

Other Eligibility Criteria

Eligibility criteria will include.

• Studies examining digital health, including AI, used by or for CHWs, as defined above;

• Studies conducted in LMICs; and

• Efficiency, as defined under Types of Interventions and Problems described above.

Types of Settings

Inclusion criteria will include studies conducted in LMIC community health settings. Studies conducted in high-income settings will be excluded during screening from the final search results but may be referenced for the Discussion section of the final EGM for contextual or comparative purposes.

Search Methods and Sources

Relevant studies will be identified through electronic searches of bibliographic databases, research networks, government policy databanks, and internet search engines. We will also use the ChatGPT platform as a supplementary tool to support the preliminary identification of relevant studies.

The searches will include studies published from 2005 onward. Results published in English, Portuguese, Spanish, and German (languages spoken by the reviewers) will be reviewed. The bibliographies of relevant reviews and included studies will be searched to identify additional references. We will conduct forward citation searching using Google Scholar, as this database produces results comparable to those of other search engines. We will also use ChatGPT as a supplementary tool to support database searches. Any results generated through AI-assisted searching will be independently reviewed for accuracy and eligibility.

Search terms will be developed based on terminology representative of implementation and dissemination research and will incorporate search filters used in previous reviews. Targeted Google searches will also be informed by the search terms used for electronic database searches. All results generated through AI-assisted methods will be independently verified and reviewed for eligibility.

Any changes in eligibility criteria will be agreed prospectively among members of the review team. Such changes will be documented and reported as deviations from the protocol in the final manuscript. In the event of changes in eligibility criteria, citations will be re-screened.

We will use EPPI-Reviewer software for screening.

Illustrative Search Strategy for PubMed

The following illustrative search strategy for PubMed was reproduced from Ballard et al. (2023):

Digital OR ICT OR Information and Communications Tech* OR Artificial Intelligence AND (Community health worker OR Health promoter OR Health educator OR Community health extension worker OR Lady health worker OR Health coach OR Community health advisor OR Family advocate OR Outreach worker OR Peer counsellor OR Patient navigator OR Health interpreter OR Public health aide OR Community Health Agents OR Community Health Assistant OR Maternal Health Worker OR Community Nutrition Worker OR Maternal & Child Health Promotion Workers OR Community-based Worker OR Community-based Health Worker OR Maternal Child Health Worker OR Nutrition Worker OR Mental Health Worker OR Postnatal Support Worker OR Community-based Skilled Birth Attendant OR Lay health worker OR Volunteer health worker OR Village health worker OR Village Malaria Worker OR Female Community Health Volunteer OR Voluntary Malaria Worker OR Nutrition Volunteer OR Community Health Volunteer OR Village Health Guide OR Community Drug Distributor OR Village Health Helper OR Mother Coordinator OR Village Drug-Kit Manager OR Community Reproductive Health Worker OR Lay Health Visitor OR Community Volunteer OR Community Health Advocate OR Community Health Aide OR Village Health Promoter OR Rural Health Worker OR Traditional Midwife OR Community Volunteer OR Lay Counselor OR Volunteer Counselor OR Volunteer Peer Counselor OR Peer Support Worker OR Shasthya Sebika OR Shasthya kormi OR Agente Comunitario de Salud OR Saksham Sahaya OR Visitadora OR Anganwadi Workers OR Promotoras de Salud OR Raedat OR Accompagnateur OR Behvarz OR Kader Posyandu OR Brigadistas OR Colaborador Voluntario OR Dai OR Bidan Kampong OR Dayas OR Doot OR Family Welfare Assistant* OR Health Assistant* OR Community Health Care Provider* OR Health Extension Worker* OR Women s Development Army OR Community Health Officer* OR Guardianes de Salud OR Accredited Social Health Activist OR Accredited Social Kader* OR Moraghebe-salamat* OR Agents Communautaires de Nutrition OR Agents Communautaires OR Health Surveillance Assistant* OR Agentes Polivalentes Elementares OR Malaria Volunteer* OR TB Volunteer* OR Village Health Worker* OR Maternal Health Worker* OR Trained TBAs OR Agents de Sant Communautaire OR Relais Volunteer* OR Volunteer Village Health Worker* OR Village Health Volunteer* OR Village Health Team member).

The above illustrative search strategy focuses on CHWs and digital technology. The screening process will filter out studies not conducted in LMICs.

Electronic Searches

Bibliographic databases to be searched:

• Association for Computing Machinery (ACM) Digital Library

• African Journals Online (AJOL)

• Business Source Complete

• Campbell Collaboration Library

• Centre for Reviews and Dissemination Databases

• Community Health Impact Coalition

• CiteSeerX

• ClinicalTrial.gov

• Cochrane Library

• EBSCO Education Resources Information Center (ERIC, via ProQuest)

• EPPI-Centre Systematic Reviews

• Google Scholar

• Ideas/Economist online

• IEEE

• International Bibliography of the Social Sciences

• International Initiative for Impact Evaluation (3ie)

• Journal of Social Work Practice LEXIS/NEXIS (including law review articles)

• LILACS (Latin American and Caribbean Health Sciences Literature)

• MEDLINE

• Mendeley

• ProQuest dissertations & theses A&I

• PsychARTICLES

• PsycINFO PubMed

• SciELO

• ScienceOpen

• Social Care Online

• Social Science Citation Abstract

• Social Science Research Network (SSRN)

• SocIndex

• SpringerLink

Searching Other Resources

We will review relevant grey literature using a separate search strategy that includes the following elements.

• Searches of grey literature databases such as ProQuest Dissertations and Theses, TROVE, and OpenGrey;

• Targeted Google searches based on the search terms used for electronic databases; and

• Searches conducted using ChatGPT as a supplementary tool to identify potentially relevant sources.

Copies of relevant documents from internet-based sources will be retained, including materials identified through ChatGPT-assisted searches. We will record the exact URL and date of access.

Snowballing

Reference lists of included studies and relevant reviews will be searched to identify additional relevant literature.

Personal Contacts

Personal contacts with national and international researchers will be used to identify unpublished reports and ongoing studies. These contacts include the WHO’s Health Workforce Unit and other WHO units, as well as national and international implementing organizations.

Analysis and Presentation

Report Structure

The report will be structured primarily according to the dimensions contained in the coding sheet and described above under Dimensions.

Multiple publications arising from the same underlying study will be identified and linked where appropriate.

Data Collection and Analysis

Screening and Study Selection

Two independent reviewers will screen study titles and abstracts independently. Disagreements between reviewers will be resolved through discussion and consensus. Potentially eligible studies will then be retrieved in full text, and full-text articles will be reviewed for eligibility by two independent reviewers. Disagreements at this stage will again be resolved through discussion and consensus.

If eligibility cannot be determined due to missing information in a report, we will contact the study authors for clarification. The completed review will include a table of studies excluded at the full-text screening stage, together with the rationale for each exclusion decision. We will also include a PRISMA flow diagram to document the screening process and results.

Data Extraction and Management

Two to three review authors, unblinded to author or journal information, will independently extract information from the included studies. This information will be recorded in a data extraction form that will be piloted before the review is initiated. Discrepancies between reviewers regarding data extraction will be resolved by consensus or, if required, by a third reviewer. One reviewer will transcribe information from the data extraction forms into study tables, and data transcription will be checked by a second reviewer.

In addition to standard coding categories, such as year of study, setting and other contextual features, target population(s), study method(s), sample size, and outcomes, we will include categories related to the classification of digital tools (e.g., mobile phone, tablet, web-based computer, AI-enabled system), characteristics of digital users, including gender and age categories, connectivity and other enabling conditions for digital tool use (e.g., literacy), and type of intervention, based on WHO classifications (World Health Organization, 2023).

Data extraction and coding will be conducted using EPPI-Reviewer (EPPI-Centre, University College London), the same platform used for screening. This software allows for structured data entry, double coding, and audit trails to support consistency and transparency. Where appropriate, we may employ AI-assisted extraction functions available within EPPI-Reviewer to support the identification and classification of study characteristics and intervention components. All AI-generated suggestions will be manually verified by reviewers before inclusion to maintain methodological rigor.

To the extent that AI tools are used for evidence synthesis or related review tasks, we will follow the Responsible AI in Evidence Synthesis (RAISE) guidelines and recommendations on the responsible use of AI (Thomas et al., 2024). We will fully and transparently document how any AI use is validated to support reliability and accuracy and to minimize risks of bias, hallucinations, or other risks typically associated with AI. We will also ensure that PRISMA and other relevant evidence synthesis standards are followed. Reference will be made to current literature outlining AI validation processes (see, e.g., Myllyaho et al., 2021).

AI-powered tools that may be considered for supporting literature review tasks include.

• MQ Library Research Assistant;

• Research Rabbit;

• scite;

• Elicit;

• SciSpace/Typeset;

• Consensus;

• Connected Papers; and

• ChatGPT.

We will validate any use of AI by:

• Checking the quality of source data; and

• Re-running analyses where appropriate to assess consistency of results.

Risk of Bias and Quality Assessment

Risk of Bias

Risk of bias will be assessed independently by two reviewers using the tool described in the Cochrane Handbook for Systematic Reviews of Interventions (Higgins et al., 2024). Judgments will be justified using information reported in the studies or related documents. If such documents are not publicly available, this will be explicitly stated. If required, a third reviewer will adjudicate discrepancies that cannot be resolved through consensus.

Unit of Analysis Issues

We will treat studies with multiple reports as a single study. This includes studies included in more than one systematic review.

Dealing With Missing Data

Situations involving missing data will be addressed on a case-by-case basis. Where relevant data are not reported, we will contact study authors to request clarification or additional information.

Quality of Evidence Assessment of Reviews

The AMSTAR 2 tool (Shea et al., 2017) will be used to assess the quality of reviews and risk of bias. Quality assessments will be conducted independently by two researchers, with discrepancies resolved through discussion.

Quality of Evidence Assessment of Studies

Quality of evidence for randomized controlled trials (RCTs) will be assessed using Cochrane’s Risk of Bias Tool. Qualitative evidence will be assessed using the JBI Critical Appraisal Tool for Qualitative Evidence (Aromataris & Munn, 2020). Two researchers will conduct quality assessments independently, with discrepancies resolved through discussion.

Data Synthesis

As this review will involve the development of an EGM, we will not perform the types of statistical analyses typically associated with systematic reviews. EGMs differ from systematic reviews in that they can identify clusters of evidence that may subsequently be analyzed through meta-analysis, with each cluster representing a potential systematic review in itself (Philbrick et al., 2022, citing Saran & White, 2018).

Subgroup Analysis and Investigation of Heterogeneity

Subgroup analyses will focus on: (1) types of digital interventions, with reference to WHO digital classifications (World Health Organization, 2023); (2) interventions incorporating AI; (3) users of digital interventions; (4) regions in which interventions are implemented; and (5) intervention outcomes.

Treatment of Qualitative Research

We intend to include qualitative research in this review. For assessing qualitative evidence, we will utilize the GRADE-CERQual tool.

Summary of Findings and Assessment of the Certainty of the Evidence

A Summary of Findings will be included in the final EGM and will be based on the evidence and results generated according to the categories outlined in the final coding sheet. At this time, we do not intend to include a formal assessment of certainty of evidence in the final EGM.

Supplemental Material

Supplemental Material - PROTOCOL: Artificial Intelligence and Other Digital Tools Used by, and for Community Health Workers (CHWs) in Low and Middle-Income Countries (LMICs) to Improve Outcomes and Increase Effectiveness: An Evidence and Gap Map

Supplemental Material for PROTOCOL: Artificial Intelligence and Other Digital Tools Used by, and for Community Health Workers (CHWs) in Low and Middle-Income Countries (LMICs) to Improve Outcomes and Increase Effectiveness: An Evidence and Gap Map by William C. Philbrick, Jacob Milnor, Feven Tassaw Mekuria, Perpetua Mbachu, Patricia Mechael, Brian Ssennoga and Zeus Aranda in Campbell Systematic Reviews.

Supplemental Material

Supplemental Material - PROTOCOL: Artificial Intelligence and Other Digital Tools Used by, and for Community Health Workers (CHWs) in Low and Middle-Income Countries (LMICs) to Improve Outcomes and Increase Effectiveness: An Evidence and Gap Map

Supplemental Material

Supplemental Material - PROTOCOL: Artificial Intelligence and Other Digital Tools Used by, and for Community Health Workers (CHWs) in Low and Middle-Income Countries (LMICs) to Improve Outcomes and Increase Effectiveness: An Evidence and Gap Map

Supplemental Material

Supplemental Material - PROTOCOL: Artificial Intelligence and Other Digital Tools Used by, and for Community Health Workers (CHWs) in Low and Middle-Income Countries (LMICs) to Improve Outcomes and Increase Effectiveness: An Evidence and Gap Map

Supplemental Material

Supplemental Material - PROTOCOL: Artificial Intelligence and Other Digital Tools Used by, and for Community Health Workers (CHWs) in Low and Middle-Income Countries (LMICs) to Improve Outcomes and Increase Effectiveness: An Evidence and Gap Map

• Content: Zeus Aranda, Perpetua Mbachu, Brian Ssennoga, Feven Mekuria, William Philbrick, and Patricia Mechael;

• EGM methods: William Philbrick and Zeus Aranda

• Statistical analysis: William Philbrick; and

• Information retrieval: William Philbrick and Zeus Aranda

Footnotes

Acknowledgements

While we have received support from a number of individuals and organizations, we wish to acknowledge in particular the following individuals and organizations who contributed to the development of this protocol: Dr. James O'Donovan, Director of Research, Community Health Impact Coalition; Nicholas Gordon, Director of Digital Health, Last Mile Health; Dr. Tamara Lotfi, Editor, Campbell Systematic Reviews; Audrey Portes, Editor, 3ie International Initiative for Impact Evaluation, Campbell Collaboration International Development Coordinating Group; Gulnaz Uzakbayeva and Diana Córdova-Arauz, Editorial Assistants, International Development Coordinating Group; Benita Rowe, Independent Consultant; and Joyce Sepenoo, Senior Director, Health Equity and Rights, CARE USA.

Author Contributions

Initial recruitment of authors for this EGM was conducted through the Community Health Impact Coalition (CHIC), a network of CHW interest holders that conducts research on behalf of CHWs (Community Health Impact Coalition, 2025). For purposes of conducting the EGM, the authors will have the following responsibilities:

• Content: Zeus Aranda, Perpetua Mbachu, Brian Ssennoga, Feven Mekuria, William Philbrick, and Patricia Mechael;

• EGM methods: William Philbrick and Zeus Aranda

• Statistical analysis: William Philbrick; and

• Information retrieval: William Philbrick and Zeus Aranda.

Funding

The authors received no financial support for the research, authorship, and/or publication of this article.

Declaration of conflicting interests

The authors declare no known conflicts of interest with respect to the research, authorship, and/or publication of this article. William Philbrick is a co-founder of an AI health start-up company that, as of the date of development of this protocol, has not launched any products.

Methods for Mapping

We plan to use EPPI-Reviewer 6 (or any later version) to develop the EGM.

Preliminary Timeframe

Depending upon the publication of this protocol, the anticipated date for submission of the draft EGM is September 1, 2026.

Plans for Updating the EGM

Plans for updating this EGM will depend upon the availability of funding. We are exploring the potential use of an AI-supported living evidence approach to enable continuous updating of the EGM. Implementation of such an approach will depend on both the availability of funding and appropriate hosting arrangements.

Supplemental Material

Supplemental Material for this article is available online.

References

Ahmed

Chase

L. E.

Wagnild

Akhter

Sturridge

Clarke

Chowdhary

Mukami

Kasim

Hampshire

(2022). Community health workers and health equity in low- and middle-income countries: Systematic review and recommendations for policy and practice. International Journal for Equity in Health, 21(1), 49. https://doi.org/10.1186/s12939-021-01615-y

Amisha

Malik

Pathania

Rathaur

V. K.

(2019). Overview of artificial intelligence in medicine. Journal of Family Medicine and Primary Care, 8(7), 2328–2331. https://doi.org/10.4103/jfmpc.jfmpc_440_19

Amoakoh-Coleman

Borgstein

A. B.

Sondaal

S. F.

Grobbee

D. E.

Miltenburg

A. S.

Verwijs

Ansah

E. K.

Browne

J. L.

Klipstein-Grobusch

(2016). Effectiveness of mHealth interventions targeting health care workers to improve pregnancy outcomes in low- and middle-income countries: A systematic review. Journal of Medical Internet Research, 18(8), e226. https://doi.org/10.2196/jmir.5533

Anderson

E. J.

McClelland

Meyer Krause

Krause

K. C.

Garcia

D. O.

Koss

M. P.

(2019). Web-based and mHealth interventions for intimate partner violence prevention: A systematic review protocol. BMJ Open, 9(8), e029880. https://doi.org/10.1136/bmjopen-2019-029880

Aranda

Vázquez

Gopaluni

Martínez

Ramírez

Jiménez

Bernal

Rodríguez

A. L.

Chacón

Vargas

Fulcher

I. R.

Barnhart

D. A.

Partners In Health Cross-Site COVID-19 Cohort Technical Working Group . (2024). Evaluation of the implementation of a community health worker-led COVID-19 contact tracing intervention in Chiapas, Mexico. BMC Health Services Research, 24(1), 97. https://doi.org/10.1186/s12913-024-10590-3

Aromataris

Munn

(Eds.), (2020). JBI manual for evidence synthesis. JBI. https://doi.org/10.46658/JBIMES-20-01

Ballard

Olaniran

Iberico

M. M.

Rogers

Thapa

Cook

Aranda

French

Olsen

H. E.

Haughton

Lassala

Carpenter Westgate

Malitoni

Juma

Perry

H. B.

(2023). Labour conditions in dual cadre community health worker programmes: A systematic review. Lancet Global Health, 11(Supplement), e1598–e1608. https://doi.org/10.1016/S2214-109X(23)00357-1

Baxter

M. S.

White

Lahti

Murto

Evans

(2021). Machine learning in a time of COVID-19: Can machine learning support community health workers in low- and middle-income countries? Journal of Global Health, 11(3), 03017. https://doi.org/10.7189/jogh.11.03017

Borges do Nascimento

I. J.

Abdulazeem

H. M.

Vasanthan

L. T.

Martinez

E. Z.

Zucoloto

M. L.

Østengaard

Azzopardi-Muscat

Zapata

Novillo-Ortiz

(2023). The global effect of digital health technologies on health workers’ competencies and health workplace: An umbrella review. The Lancet Digital Health, 5(8), e534–e544. https://doi.org/10.1016/S2589-7500(23)00092-4

10.

Braun

Catalani

Wimbush

Israelski

(2013). Community health workers and Mobile technology: A systematic review of the literature. PLoS One, 8(6), e65772. https://doi.org/10.1371/journal.pone.0065772

11.

Ciecierski-Holmes

Singh

Axt

Brenner

Barteit

(2022). Artificial intelligence for strengthening healthcare systems in low- and middle-income countries: A systematic scoping review. Npj Digital Medicine, 5(1), 162. https://doi.org/10.1038/s41746-022-00700-0

12.

Community Health Impact Coalition . (2025). Research. https://joinchic.org/what-we-do/research/

13.

Crigler

Hill

Furth

Bjerregaard

(2013). Community health worker assessment and improvement matrix (CHW AIM): A toolkit for improving community health worker programs and services (revised version). University Research Co., LLC. https://pdf.usaid.gov/pdf_docs/PA00JV46.pdf

14.

De Leeuw

(2019). A systematic review of digital health tools used for decision support by frontline health workers in low- and middle-income countries. https://hdl.handle.net/11427/31402

15.

Duke Global Health Innovation Center . (2024). Transforming healthcare: Unleashing the power of AI for frontline health workers. https://dukeghic.org/2024/03/21/transforming-healthcare-unleashing-the-power-of-ai-for-frontline-workers/

16.

Feroz

A. S.

Khoja

Saleem

(2021a). Equipping community health workers with digital tools for pandemic response in low- and middle-income countries. Archives of Public Health, 79(1), 1. https://doi.org/10.1186/s13690-020-00513-z

17.

Feroz

A. S.

Valliani

Khwaja

Karim

(2021b). Exploring digital health interventions to support community health workers in low- and middle-income countries during COVID-19: A scoping review protocol. BMJ Open, 11(9), e053871. https://doi.org/10.1136/bmjopen-2021-053871

18.

Fischer

(2024). AI and community health workers. https://globalaccessaction.org/projects/

19.

Fitzpatrick-Lewis

Ciliska

Thomas

(2009). The methods for the synthesis of studies without control groups. National Collaborating Centre for Methods and Tools. https://www.nccmt.ca/pubs/non-RCT2_EN.pdf

20.

Greenhalgh

Robert

Macfarlane

Bate

Kyriakidou

Peacock

(2005). Diffusion of innovations in service organizations: Systematic review. Social Science & Medicine, 61(2), 417–430. https://doi.org/10.1016/j.socscimed.2004.12.001

21.

Hasan

M. Z.

Dinsa

G. D.

Berman

(2021). A practical measure of health facility efficiency: An innovation in the application of routine health information to determine health worker productivity in Ethiopia. Human Resources for Health, 19(1), 96. https://doi.org/10.1186/s12960-021-00636-6

22.

Helldén

Tesfaye

Gachen

Lindstrand

Källander

(2023). Digital health funding for COVID-19 vaccine deployment across four major donor agencies. The Lancet Digital Health, 5(9), e627–e631. https://doi.org/10.1016/S2589-7500(23)00134-6

23.

Higgins

J. P. T.

Thomas

Chandler

Cumpston

Page

M. J.

Welch

V. A.

(Eds.), (2024). Cochrane handbook for systematic reviews of interventions. Cochrane. https://www.cochrane.org/handbook

24.

Igwama

G. T.

Nwankwo

E. J.

Emeihe

E. V.

Ajegbile

M. D.

(2024). The role of community health workers in implementing AI-based health solutions in rural areas. International Journal of Biology and Pharmacy Research Updates. https://orionjournals.com/ijbpru/content/role-community-health-workers-implementing-ai-based-health-solutions-rural-areas

25.

Islam

Mayeesha

T. T.

Ahmed

(2024). Know your users: Towards explainable AI in Bangladesh. In Companion of the 2024 ACM international joint conference on pervasive and ubiquitous computing (ubiComp companion ’24) (pp. 889–892). Association for Computing Machinery. https://doi.org/10.1145/3675094.3679002

26.

Last Mile Health . (2024). https://lastmilehealth.org/

27.

(2025). Evolution of artificial intelligence in medical education from 2000 to 2024: Bibliometric analysis. Interactive Journal of Medical Research, 14(1), e63775. https://doi.org/10.2196/63775

28.

Living Goods, Health.Enabled, Johnson & Johnson . (2019). Digital health tools from community health worker programs maturity model and toolkit. https://livinggoods.org/what-we-do/digitalhealth/

29.

Lotito

(2025). Foreign Assistance Cuts Hamper Community Health Workers and Patients. Forbes. https://www.forbes.com/sites/jenniferlotito/2025/04/16/foreign-assistance-cuts-hamper-community-health-workers-and-patients/

30.

MacPherson

Webb

E. L.

Kamchedzera

Joekes

Mjoli

Lalloo

D. G.

Divala

T. H.

Choko

A. T.

Burke

R. M.

Maheswaran

Pai

Squire

S. B.

Nliwasa

Corbett

E. L.

(2021). Computer-aided X-ray screening for tuberculosis and HIV testing among adults with cough in Malawi (the PROSPECT study): A randomised trial and cost-effectiveness analysis. PLoS Medicine, 18(9), e1003752. https://doi.org/10.1371/journal.pmed.1003752

31.

Matthias

(2022). Strategic and operational pathway to strengthen the effectiveness of frontline health workers at scale through a global mHealth tool and social business. Springer Nature Link.

32.

Menon

Shimelash

Rutunda

Nshimiyimana

Archer

Emmanuel-Fabula

Berhe

D. F.

Gill

Hezagira

Remera

Riley

Wong

Denniston

A. K.

Mateen

B. A.

Liu

(2025). Assessing the potential utility of large language models for assisting community health workers: Protocol for a prospective observational study in Rwanda. BMJ Open, 15(10), e110927. https://doi.org/10.1136/bmjopen-2025-110927

33.

Myllyaho

Raatikainen

Männistö

Mikkonen

Nurminen

J. K.

(2021). Systematic literature review of validation methods for AI systems. ArXiv, abs/2107.12190.

34.

Norris

Atkins

(2005). Challenges in using nonrandomized studies in systematic reviews of treatment interventions. Annals of Internal Medicine, 142(12_Part_2), 1112–1119. https://doi.org/10.7326/0003-4819-142-12_part_2-200506211-00011

35.

Odendaal

W. A.

Anstey Watkins

Leon

Goudge

Griffiths

Tomlinson

Daniels

(2020). Health workers’ perceptions and experiences of using mHealth technologies to deliver primary healthcare services: A qualitative evidence synthesis. Cochrane Database of Systematic Reviews, 3(3), CD011942. Available at: https://doi.org/10.1002/14651858.CD011942.pub2

36.

O'Donovan

Kahn

MacRae

Namanda

A. S.

Hamala

Kabali

Geniets

Lakati

Mbae

S. M.

Winters

(2022). Analysing 3429 digital supervisory interactions between community health workers in Uganda and Kenya: The development, testing and validation of an open access predictive machine learning web app. Human Resources for Health, 20(1), 26. https://doi.org/10.1186/s12960-021-00699-5

37.

Ogolodom

M. P.

Mbaba

A. N.

Johnson

Chiegwu

H. U.

Ordu

K. S.

Okeji

M. C.

Alazigha

Balogun

E. O.

Bakre

A. F. K.

Joseph

D. Z.

Dambele

M. Y.

Nyenke

C. U.

Madume

A. K.

Jeremiah

C. U.

Egop

B. E.

Ochong

A. D.

Nwodo

V. K.

(2023). Knowledge and perception of healthcare workers towards the adoption of artificial intelligence in healthcare service delivery in Nigeria. AG Salud, 1(1), 16. https://doi.org/10.62486/agsalud202316

38.

Paduano

Incerti

Borsari

Benski

A. C.

Ernest

Mwampagatwa

Lilungulu

Masoi

Bargellini

Stornelli

Stancanelli

Borella

Rweyemamu

M. A.

(2022). Use of a mHealth system to improve antenatal care in low and lower-middle income countries: Report on patients and healthcare workers’ acceptability in Tanzania. International Journal of Environmental Research and Public Health, 19(22), 15342. https://doi.org/10.3390/ijerph192215342

39.

Palmer

M. J.

Henschke

Bergman

Villanueva

Maayan

Tamrat

Mehl

G. L.

Glenton

Lewin

Fønhus

M. S.

Free

(2020). Targeted client communication via Mobile devices for improving maternal, neonatal, and child health. Cochrane Database of Systematic Reviews, 8(8), CD013679. https://doi.org/10.1002/14651858.CD013679

40.

PATH . (2025a). New study examines how generative AI can assist community health workers in Rwanda. PATH Media Center. https://www.path.org/our-impact/media-center/new-study-examines-how-generative-ai-can-assist-community-health-workers-in-rwanda/

41.

PATH . (2025b). PATH launches AI clinical trial with penda health. PATH Media Center. https://www.path.org/our-impact/media-center/path-launches-artifical-intelligence-clinical-trial/

42.

Peters

R. P. H.

Grinsztejn

Celum

Mayer

K. H.

Molina

J. M.

Delany-Moretlwe

Ngure

Torres-Rueda

Guy

R. J.

Reiss

Ferrand

R. A.

Kittiyaowamarn

Bekker

L. G.

(2025). Innovations in the biomedical prevention, diagnosis, and service delivery of HIV and other sexually transmitted infections. The Lancet, 406(10515), 2133–2151. https://doi.org/10.1016/S0140-6736(25)00983-3

43.

Philbrick

Milnor

Deshmukh

Mechael

(2022). Information and communications technology use to prevent and respond to sexual and gender-based violence in low- and middle-income countries: An evidence and gap map. Campbell Systematic Reviews, 18(4), e1277. https://doi.org/10.1002/cl2.1277

44.

Rodrigues

S. M.

Kanduri

Nyamathi

Dutt

Khargonekar

Rahmani

A. M.

(2022). Digital health-enabled community-centered care: Scalable model to empower future community health workers using human-in-the-loop artificial intelligence. JMIR formative research, 6(4), e29535. https://doi.org/10.2196/29535

45.

Saran

White

(2018). Evidence and gap maps: A comparison of different approaches. Campbell Systematic Reviews, 14(1), 1–38. https://doi.org/10.4073/cmdp.2018.2

46.

Shea

B. J.

Reeves

B. C.

Wells

Thuku

Hamel

Moran

Moher

Tugwell

Welch

Kristjansson

Henry

D. A.

(2017). Amstar 2: A critical appraisal tool for systematic reviews. BMJ, 358(1), j4008. https://doi.org/10.1136/bmj.j4008

47.

Shi

Bendig

Vollmar

H. C.

Rasche

(2023). Mapping the bibliometrics landscape of AI in medicine: Methodological study. Journal of Medical Internet Research, 25(4), e45815. https://doi.org/10.2196/45815

48.

Solano-Kamaiko

Mishra

Dell

Vashistha

(2024). Explorable explainable AI: Improving AI understanding for community health workers in India. In Proceedings of the CHI conference on human factors in computing systems (pp. 1–21). Association for Computing Machinery. https://doi.org/10.1145/3613904.3642733

49.

Thomas

Flemyng

Noel-Storr

Moy

Marshall

Hajji

Jordan

Aromataris

Mheissen

Clark

A. J.

Jamioło

Saran

Walker

V. R.

Angrish

Macura

Spijker

Kusa

Wojciech

Chi

Fletcher

(2024). Responsible AI in evidence synthesis (RAISE): Guidance and recommendations. https://osf.io/cn7x4

50.

Wankhede

P. R.

Bhuyar

Zanwar

Pawar

Jadhav

M. R.

Gandhewar

Kulkarni

M. B.

Bhaiyya

Haick

(2025). Artificial intelligence for noninvasive health diagnostics. ACS Sensors, 10(11), 8217–8255. https://doi.org/10.1021/acssensors.5c03171

51.

Whidden

Kayentao

Liu

J. X.

Lee

Keita

Diakité

Keita

Diarra

Edwards

Yembrick

Holeman

Samaké

Plea

Coumaré

Johnson

A. D.

(2018). Improving community health worker performance by using a personalised feedback dashboard for supervision: A randomised controlled trial. Journal of Global Health, 8(2), 020418. https://doi.org/10.7189/jogh.08.020418

52.

White

Albers

Gaarder

Kornør

Littell

Marshall

Mathew

Pigott

Snilstveit

Waddington

Welch

(2020). Guidance for producing a campbell evidence and gap map. Campbell Systematic Reviews, 16(4), e1125. https://doi.org/10.1002/cl2.1125

53.

Whittemore

Knafl

(2005). The integrative review: Updated methodology. Journal of Advanced Nursing, 52(5), 546–553. https://doi.org/10.1111/j.1365-2648.2005.03621.x

54.

World Economic Forum & Digital Health Action Alliance . (2023). Wiring the community health worker: A winning strategy for NCD care. https://www3.weforum.org/docs/WEF_Wiring_the_Community_Health_Worker_2023.pdf

55.

World Health Organization . (2006). The world health report 2006: Working together for health. WHO. https://www.who.int/whr/2006/en/

56.

World Health Organization . (2019). WHO guideline: Recommendations on digital interventions for health system strengthening. WHO. https://www.who.int/publications

57.

World Health Organization . (2020). What Do We Know About Community Health Workers? A Systematic Review of Existing Reviews. WHO. https://www.who.int/publications/i/item/what-do-we-know-about-community-health-workers-a-systematic-review-of-existing-reviews

58.

World Health Organization . (2023). Classification of digital interventions, services and applications in health (2nd ed.). https://iris.who.int

59.

World Health Organization . (2024). Ethics and governance of artificial intelligence for health. WHO. https://www.who.int

60.

Xie

Zhai

(2025). Evolution of artificial intelligence in healthcare: A 30-year bibliometric study. Frontiers in Medicine, 12(7), 1505692. https://doi.org/10.3389/fmed.2024.1505692

61.

Zarska

Avgar

A. C.

Sterling

M. R.

(2021). Relationship between working conditions, worker outcomes, and patient care: A theoretical model for frontline health care workers. American Journal of Medical Quality: The Official Journal of the American College of Medical Quality, 36(6), 429–440. https://doi.org/10.1097/01.JMQ.0000735508.08292.73

62.

Zaza

Wright-De Aguero

L. K.

Briss

P. A.

Truman

B. I.

Hopkins

D. P.

Hennessy

M. H.

, et al. (2000). Data collection instrument and procedure for systematic reviews in the guide to community preventive services. American Journal of Preventive Medicine, 18, 44–74. https://doi.org/10.1016/S0749-3797(99)00122-1

Supplementary Material

Please find the following supplemental material available below.

For Open Access articles published under a Creative Commons License, all supplemental material carries the same license as the article it is associated with.

For non-Open Access articles published, all supplemental material carries a non-exclusive license, and permission requests for re-use of supplemental material or any part of supplemental material shall be sent directly to the copyright owner as specified in the copyright notice associated with the article.

0.37 MB

0.06 MB

0.12 MB

0.00 MB

0.06 MB