Artificial Intelligence in Health Promotion and Disease Reduction: Rapid Review

Introduction

As per the World Health Organization, noncommunicable diseases are responsible for 41 million deaths annually, constituting 71% of the total mortality [1-3]. These diseases are characterized by prolonged duration, risks of complications, and long-term treatment, significantly impacting people’s health and quality of life [4]. Behavioral risk factors such as smoking, inadequate nutrition, alcohol consumption, and lack of physical activity [5] contribute to the development of these conditions [6]. Chronic diseases have emerged as a significant challenge to global health, which underscores the crucial need for a targeted approach in health policy to address this issue [6,7]. Actively promoting the prevention and management of chronic diseases is essential to achieving broader health care goals, such as (1) improving the patient experience, (2) improving population health, (3) reducing costs, (4) promoting equity, (5) supporting health care providers, and (6) enhancing system efficiency, referred to as the “sextuple aim” [8].

Most chronic diseases can be prevented or improved through health promotion strategies [9], thereby mitigating risk factors through the dissemination of lifestyle information and advocacy of healthy habits [10]. Successful health promotion programs are those that focus on imparting knowledge, influencing attitudes, and changing behaviors [11]. Health promotion strategies, such as enhancing public awareness, changing behavior, creating supportive environments, and promoting healthy eating and active lifestyles, are key to preventing chronic diseases [12,13]. In this context, self-management is a crucial approach to managing chronic conditions, emphasizing patient responsibility and problem-solving [14,15]. Health promotion refers to proactive efforts aimed at empowering individuals and communities to adopt healthier lifestyles, prevent diseases, and enhance overall well-being [16,17]. Disease reduction, on the other hand, focuses on minimizing the incidence, severity, and long-term impacts of chronic conditions through targeted interventions and preventive strategies [18].

Recent advancements in artificial intelligence (AI) can be widely used in the medical field to improve patient care [19] and can also be effective in encouraging healthy behaviors and lifestyle changes [20]. AI, as a branch of computer science, simulates human cognition to perform tasks such as reasoning, learning, and decision-making [21,22]. Its core techniques, such as machine learning, deep learning, natural language processing, knowledge representation, and reasoning, have been deployed in health research areas [23], increasing its popularity in the medical community [22-25]. It has offered significant potential for public health practitioners and policymakers by supporting context-specific, data-driven interventions [26]. AI’s ability to reason, act, and adapt based on data is particularly valuable in the early prevention of diseases [27], treatment optimization [24], and delivering personalized, engaging self-care solutions [27,28].

The positive influence of AI in health care can include significant transformations in health promotion and disease prevention. Early disease detection and prevention can motivate the population to adopt healthy eating, lifestyle, and exercise habits [29]; improve patient outcomes by lowering disease severity; and reduce the economic burden on health systems [30]. However, the acceptance and application of these technologies for health, especially in certain populations such as older adults, are much slower than their emergence and development [31]. This highlights the importance of identifying practical applications of AI in promoting healthy lifestyles to reduce the risk of diseases, as well as the essential conditions for its effective implementation.

Evidence of AI and machine learning applications for health promotion has been synthesized specifically for promoting health behavior change, physical activity, and healthy eating [20,32,33]. However, most existing research has primarily focused on AI applications in clinical settings and individualized interventions, with limited attention to its role in community-based and nonclinical health promotion programs. AI-driven technologies, such as digital health assistants, mobile interventions, and community engagement platforms, have the potential to enhance public health initiatives by increasing the accessibility, scalability, and personalization of preventive strategies [34,35]. Despite this potential, there remains a lack of synthesized knowledge on the implementation of AI in broader, nonclinical contexts, which limits its integration into public health policies and community-wide interventions. This gap in evidence further restricts the ability to apply AI-based solutions effectively in real-world health promotion and disease prevention strategies.

This review thus aims to: (1) identify AI-based initiatives and their purposes; (2) synthesize their impacts on health promotion domains; and (3) categorize the challenges, benefits, and suggestions regarding successful AI implementation in health promotion and disease reduction, using a Strengths, Weaknesses, Opportunities, and Threats (SWOT) matrix [36].

Methods

Overview

We conducted a rapid knowledge synthesis following the Cochrane Collaboration guidance for rapid reviews [37]. The results are presented based on the PRISMA (Preferred Reporting Items for Systematic Reviews and Meta-Analyses) guidelines [38,39]. We prospectively registered the protocol for this rapid review in the Open Science Framework Registries [40].

Eligibility Criteria

Our search strategy was guided by the Population, Concept, and Context framework, recommended by the Johanna Briggs Institute [41,42], to ensure a comprehensive and focused search. This framework suited our study due to its appropriateness for a broad research subject (Table 1).

Literature Search

We developed a comprehensive list of keywords, synonyms, and MeSH (Medical Subject Headings) terms related to the main concepts of our research question, guided by the Population, Concept, and Context framework. This strategy was formulated in collaboration with an information specialist (FB), aiming to capture the broad spectrum of AI applications in health promotion and disease prevention. We searched two scientific databases aligned with the subject of interest, MEDLINE (Ovid) and CINAHL, in March 2024, to locate studies published between January 1, 2019, and March 15, 2024 (Multimedia Appendix 2). No other restrictions were applied. Since the use of AI in health has been promoted in recent years, focusing on the literature of the last 5 years provides us with more up-to-date findings, as this field evolves rapidly.

Study Selection, Data Extraction, and Synthesis

We exported all citations to the Covidence web-based collaborative tool [43], where duplicates were removed manually and using the automated function. Seven reviewers (FY, FN, SO, MPG, MC, MO, MS) working in pairs performed an independent assessment of titles and abstracts, and then 5 (FY, FN, MPG, SO, ARY) reviewed full texts. Any discrepancies between reviewers during the title and abstract screening or full-text review were resolved through discussion. If consensus could not be reached, a senior reviewer (MPG) made the final decision. We performed a calibration exercise on 10 citations to ensure that the selection criteria were understood in the same way by all reviewers. The same reviewers completed data extraction in an Excel 2016 (Microsoft Corp) form, including study characteristics (eg, authors, title, year, country, design, and population data), intervention characteristics (type of AI initiatives), intervention purpose (eg, health promotion domain, impacts), and AI implementation (challenges, opportunities, and suggestions). A single reviewer extracted data, which was then confirmed by a senior reviewer (MPG). In cases where discrepancies arose during data extraction, they were discussed and resolved collectively. When needed, the senior reviewer (MPG) provided the final decision to ensure consistency in data reporting. Although initial data extraction was conducted by a single reviewer, all extracted data were independently verified by a senior reviewer to minimize bias and enhance reliability. This two-step verification approach aligns with established practices in rapid reviews [44]. We used a PRISMA 2020 flow diagram [45] to describe the study identification, screening process, and application of inclusion and exclusion criteria.

We synthesized the extracted data from study characteristics, intervention characteristics, and intervention purpose using structured narrative summaries of main themes, which gave a portrait of the current scope of available AI initiatives used in promoting healthy activities and preventing disease. We grouped the reported outcomes into three categories, namely, knowledge-cognition-behavior, health status, and process. We used thematic analysis [46] to categorize and summarize the challenges, benefits, and suggestions related to AI implementation in health promotion and disease prevention, according to a SWOT matrix.

A SWOT analysis systematically evaluates an organization’s internal strengths and weaknesses alongside external opportunities and threats within its environment. This internal analysis aims to identify the organization’s resources, capabilities, core competencies, and competitive advantages, categorized as Strengths and Weaknesses. Conversely, the external analysis assesses market opportunities and threats by examining competitor resources, the industry landscape, and broader environmental factors grouped as Opportunities and Threats. The goal of a SWOT analysis is to leverage insights about the organization’s internal and external contexts to inform strategic decision-making [36]. This matrix identifies potential tactical strategies for exploiting opportunities or mitigating threats by leveraging existing strengths and minimizing weaknesses, facilitating the development of tactical strategies from 4 distinct perspectives [36,47,48] (Figure 1).

‎

Results

Study Selection

A total of 3876 publications were retrieved, and 434 duplicates were removed both manually and automatically (by Covidence). The remaining 3442 publications were screened by independent reviewers using titles and abstracts. Among them, 58 publications were screened in full text, resulting in 23 publications describing 22 studies suitable for inclusion in this review (see Figure 2).

‎

Characteristics of the Included Studies

The included studies were published between 2019 and 2023 (see Table 2 and Figure 2), and most of them (7/22, 32%) were published in 2023. Ten studies (45.5%) were conducted in the United States, 3 (14%) in Japan, 3 (14%) in Australia, 2 (9%) in the United Kingdom, 2 (9%) in Spain, 1 (4%) in Canada, and 1 (4%) in Italy (see Figure 3). The higher representation of studies from the United States may be attributed to greater investment in AI research, a well-established digital health infrastructure, and large-scale government-supported AI initiatives, facilitating AI-driven public health applications [49,50]. The number of participants involved in each study ranged from 21 to 139,164, with participants generally being recruited voluntarily and without any specific health concerns. Of the 22 studies included in our rapid review, 11 (50%) were randomized or nonrandomized trials, 3 (14%) were cohort studies, 2 (9%) were feasibility studies, 2 (9%) were observational studies, 1 (4%) study used a mixed method research design, and 1 (4%) study used a quasi-experimental design (see Table 2).

‎

These studies evaluated the use and effectiveness of different types of AI systems (mobile apps, software, websites, platforms, devices, and chatbots) in various health promotion domains (dietary behavior, physical activity, smoking cessation, mental health, and preventive care). The most frequently used AI systems were mobile apps (14/22) and chatbots (9/22), while websites (5/22), software (2/22), and platforms (1/22) were less common. Notably, 6 of 22 studies used a combination of two types of AI systems, for instance, the combination of software and a website or a mobile app and a chatbot [53,54,56,64,65,68,71]. Regarding health promotion domains, 10 studies focused on dietary behavior, 6 related to smoking cessation, 4 focused on physical activity, 3 on mental health, and 3 on preventive care. Additionally, 4 studies incorporated multiple domains within a single AI-based intervention, integrating aspects like diet and physical activity or smoking cessation and mental health [52,64,65,72,73] (see Table 2, Table 3, and Figure 3).

Primary Outcomes Measured

The primary outcome of this review was the impact of AI initiatives on the health promotion domain. Table 4 presents an overview of health promotion domains considered in each study. Disease reduction was directly targeted in studies about the self-management of hypertension [52], diabetes [56,73], obesity [63], and both obesity and prediabetes or diabetes [62,70]. All the studies reported positive impacts of the AI-based intervention on each domain of health promotion, although 10 of them reported uncertain results, and 3 of them also reported negative results.

Specific impacts reported were grouped into 3 categories (knowledge-cognition-behavior, health status, and process), as depicted in Table 4. Smoking cessation had the most reported impact, with 4 studies showing statistically significant positive outcomes related to AI [57,60,68,69]. Physical activity and dietary behavior also showed positive impacts, with 3 [63,64,71] and 2 [64,73] studies with statistically significant positive outcomes, respectively. Regarding health status outcomes, 6 studies reported statistically significant impacts of AI interventions on weight loss [53,60,62,64,66,73], 2 on mental health dimensions [53,72], 1 on glycemic control [73], and 1 on systolic blood pressure [52]. However, some of these impacts were not sustained over time, as reported [61,67].

Process outcomes were generally assessed through descriptive methods. Most studies showed that AI-mediated interventions increased participant engagement, adherence, and satisfaction [51,53,58,61,62,68,69,71,72]. Ease of use, usefulness, and enjoyment were often reported with AI interventions [55,56,58,61,67,70,71]. However, some studies reported more negative experiences with the use of AI interventions such as chatbots [56,64,65]. Decrease in engagement over time, frustration regarding the interaction with the chatbot, and the need for human interaction were identified as potential drawbacks of AI-supported health promotion interventions [56,61,64,65,71,72].

Secondary Outcomes Measured

Using thematic analysis, we categorized challenges, benefits, and suggestions into several subgroups. Challenges are presented in 5 subcategories: Ineffectiveness, Humanity and Specialty Needs, Lack of Engagement, Technical Limitations, and Unqualified Data. Benefits are divided into Practical Impacts, Positive Encouragement, Clinical Process Improvement, and Cost Effectiveness. In addition, suggestions are categorized into 5 subgroups: Future Investigation, Stakeholder Collaboration, End User Consideration, Technical Improvement, and Resource Allocation. The details of these categories are provided in Multimedia Appendix 3.

We used a SWOT matrix to present and analyze these results. For this, we focused separately on internal factors that strengthen or weaken AI implementation in health promotion and disease reduction, as well as external factors that create opportunities and threats for this context. Therefore, in this way, these 4 dimensions can be defined as follows:

1. Strengths (S): superior skills and resources that enhance the performance of AI systems in health promotion and disease reduction.
2. Weaknesses (W): deficiencies that can hinder the success of AI implementation in health promotion and disease reduction.
3. Opportunities (O): external trends that can improve the performance of these systems in promoting health or reducing disease if leveraged.
4. Threats (T): environmental factors that may decrease the performance of these systems in promoting health or reducing disease if not addressed.

The SWOT matrix is presented in Textbox 1. Internal challenges are defined as weaknesses and external ones are defined as threats. Similarly, the internal benefits are categorized as strengths and external ones as opportunities.

Weaknesses and Threats

The most challenging factors in AI implementation in health promotion were identified as the short-term impacts of AI interventions in promoting health [54,57,66,67,72], requiring human intervention and specialization of providers [55,56,66-68], lack of participant engagement in related studies [54-56,67], technical limitations of AI-based systems (eg, lack of access to past clinical data) [57,61,62,66], and concerns about data quality [57,62]. In categorizing the challenges, deficiencies in the performance of AI initiatives and the lack of optimal evaluation of them are considered weaknesses. On the other hand, external negative impacts, such as the short-term implications of AI initiatives, the requirement for expert interpretation of the outputs, and the lack of consensus on improving health and healthy lifestyles, are seen as threats.

Strengths and Opportunities

Along with the challenges, this study also highlighted the benefits of using AI to promote health and reduce disease. These include facilitating patient access [54,68] by providing straightforward and applicable systems [54], the improvement of personalized coaching with detailed feedback, preventive care [53,54,57,62,68], encouraging people to incorporate healthy behaviors [53,55,56,62,68], clinical process and workflow improvement [55,62,66,68], and providing cost-effectiveness interventions [53,55,61,68]. When discussing the benefits, internal ones such as providing feedback [53,62,68], reducing the drawbacks of interventions [53,57,62], and using cost-sensitive algorithms [61] are considered strengths. External positive impacts, such as providing accessible interventions [53,54,56,68], motivating people to self-track and live healthy [53,55,56,62], not creating stress for providers [55], reducing the providers’ workload [55,66,68], and mitigating excess demand [53,55,68], are seen as opportunities.

OS, OW, TS, and TW Strategies

The identified recommendations include more investigation into the different aspects of AI application in health promotion, such as economic aspects [51,53,55,56,69], the collaboration between diverse stakeholders [56,57,59], considering end users’ needs and preferences in planning and designing AI-based interventions [56,57,59,64,67,72,73], enhancing technical performance and features of the AI initiatives [51,54,61,62,64,68,69], and allocating sufficient financial and human resources [64,69]. By using SWOT analysis, we display suggestions across 4 categories: OS (Opportunities, Strengths), OW (Opportunities, Weaknesses), TS (Threats, Strengths), and TW (Threats, Weaknesses) (Table 5). To explain further, using strengths such as the cost-effectiveness of AI algorithms can enhance the opportunities for exploring different aspects of AI. In addition, the benefits of saving money and reducing costs can lead to more technical advancements in AI systems, achievable through collaboration with key stakeholders. This approach can not only improve weaknesses but also open up new opportunities.

Discussion

Principal Findings

We conducted this rapid knowledge synthesis to address the growing demand for evidence on the practical implementation of AI-based initiatives in promoting health and reducing diseases. Our review highlights several benefits of integrating AI into health promotion and chronic disease self-management interventions, both of which contribute directly to disease reduction. AI interventions showed significant improvements, particularly in promoting smoking cessation, physical activity, and dietary patterns.

Our research findings align with the extensive evidence demonstrating the efficacy of AI initiatives in promoting health and reducing chronic diseases [74]. Our analysis contributes to the growing body of evidence on the efficacy of AI in enhancing self-management for individuals living with hypertension, prediabetes, diabetes, and obesity [54,56-58,61,66,69,73]. A systematic review has also demonstrated the efficacy of AI chatbots in areas such as promoting healthy behavior and lifestyles, smoking cessation, and medication adherence [20]. Several AI-driven interventions have demonstrated significant success in real-world applications. For instance, AI-powered chatbots have effectively facilitated smoking cessation by providing personalized motivation and behavioral feedback [69,75]. Additionally, mobile-based AI interventions have shown positive impacts on promoting physical activity by offering adaptive goal-setting and real-time coaching, leading to improved adherence to exercise routines [63,76]. Although both types of interventions showed promise, chatbot-based solutions appeared particularly effective for initiating behavior change, whereas mobile apps were better suited for sustaining long-term adherence. However, the heterogeneity in study designs and outcome measures prevents definitive comparisons between intervention types. These patterns indicate the potential for targeted use of specific AI modalities, but more head-to-head studies are needed to establish comparative effectiveness.

Our review also shows that AI has the potential to influence the health status of the population by aiding in weight loss and mental health; however, its unsustainability over time is underscored. In line with our findings, it is reported that AI has the potential to aid in weight loss [77] and enhance mental health [78], but its use is still undeveloped, and the ethical challenges must be addressed [77,78]. AI-based interventions also yielded significant improvements in the process outcomes, leading to increased participant engagement, adherence, and satisfaction. It is, however, important to note that there were documented instances of adverse experiences that manifested in reduced engagement over time and frustration during interaction [56,64,65]. Certain AI interventions struggled to sustain long-term behavior change, possibly due to the novelty effect or limited personalization. Some studies also reported challenges in adapting AI tools to users’ evolving needs and a lack of follow-up assessments to confirm lasting effects [20,32,79]. To address these challenges, incorporating strategies from behavioral science and user experience design, such as gamification, adaptive feedback loops, and interactive AI coaching, could improve long-term user engagement and adherence to AI-driven health interventions [80-82]. Limited direct comparisons with traditional non-AI interventions suggest that AI-powered tools may enhance personalization and accessibility in ways that conventional methods cannot, particularly in remote monitoring and behavioral coaching [83,84]. However, further research is needed to compare the long-term efficacy and user satisfaction between AI-driven and traditional health promotion strategies to determine their relative advantages and potential complementarities.

Regarding the overall experiences of using AI in health promotion, previous reviews have highlighted the impact of AI interventions on enhancing participant involvement and their high usability and acceptability in health-related activities [20,85,86]. Beyond the findings of this review, AI has significant potential to enhance health care delivery in low-resource and underserved settings [87]. AI-driven tools can bridge gaps in access by enabling remote monitoring, telehealth, and automated decision support, particularly in areas with workforce shortages. Mobile AI apps and chatbots could also improve access to evidence-based health information, helping individuals overcome barriers related to geography, cost, and availability of care [88,89]. However, the scalability of AI-based interventions in rural and low-income settings depends on several key factors, including infrastructure availability, digital literacy, and integration with existing health care workflows [90,91]. Without proactive efforts to bridge these gaps, AI could inadvertently reinforce existing health inequities. Therefore, equity-oriented implementation strategies—including localized adaptation, capacity building, and inclusive design—are essential to ensure that AI benefits are equitably distributed. Indeed, for sustainable implementation, AI solutions must be tailored to the specific needs of underserved populations, ensuring that technological advancements do not inadvertently widen health disparities [91,92]. Future initiatives should prioritize context-specific adaptations and capacity-building efforts to maximize the reach and effectiveness of AI-driven health interventions.

However, it has been discussed that AI can raise concerns about trust, responsibility, and emotional engagement [93]. These concerns are particularly relevant in mental health applications, where ensuring ethical safeguards and human oversight is essential to building trust in AI-driven interventions [94,95]. Although our review identified key ethical and technical challenges associated with AI in health promotion, additional challenges must also be considered. Data privacy, participant consent, and algorithmic bias are critical considerations in AI-driven health care. Although AI can enhance engagement and accessibility [53,62,68], concerns remain regarding data security, informed consent, and the potential misuse of sensitive health information [96,97]. Biases in training datasets may also lead to inequitable health outcomes, reinforcing disparities in underserved populations [41,98]. These biases can disproportionately affect certain demographic groups based on factors such as gender, age, socioeconomic status, and cultural background [99,100]. Addressing these challenges requires transparent data governance, adherence to regulatory frameworks (eg, General Data Protection Regulation, Health Insurance Portability and Accountability Act), and robust bias mitigation strategies [101-104]. Additionally, implementing explainable AI and strengthening cybersecurity protocols can help ensure fairness, user trust, and ethical AI integration in health care [105,106]. Furthermore, the development of inclusive smart health models, the enhancement of data diversity, the application of debiasing techniques, and the adoption of human-centered approaches that prioritize the needs of marginalized communities are recommended to mitigate these issues [99,107]. This demonstrates that the integration of AI in health care remains a nascent emerging field requiring more transparent and trustworthy implementation [108].

We also identified the related challenges, benefits, and suggestions from the included studies and analyzed them using a SWOT analysis. According to the SWOT analysis, the predominant strengths include providing supportive coaching and feedback for users, achieving a high rate of participation, and improving clinical processes. These favorable attributes of AI are consistent with other research on the benefits of allowing users to receive personalized care and guidance in managing and controlling their health status [109], and engaging more in persuading their care plan [110], all without adding extra burdens on clinical procedures [111]. The main weaknesses identified were mostly related to technical limitations, unqualified data, and human aspects. In comparison to related studies, our findings align with previous reviews, emphasizing the constraints of insufficient data, poor validation, ethical concerns, legal barriers [23,112,113], and technical obstacles [114].

The identified opportunities encompassed aspects such as patient motivation, reduction of health professionals’ workload, and interventions aimed at cost-saving. The findings are in line with previous research, underscoring the important role of AI in decreasing the workload of health professionals, addressing the workforce shortage [115], improving patient engagement and comprehension [116], and reducing health care costs [117]. The main threats were the imperative for human involvement in results interpretation and communication, along with the short-term impacts of AI-based interventions. Corresponding with our results, Udegbe’s review [118] revealed the intricacies of human-AI interaction as an important obstacle in AI implementation in health care. Furthermore, the limited generalizability, the need for robust clinical evaluation [119,120], and problems with sustainable usability and efficacy [32,33] were cited as critical challenges in implementing AI for promoting health and reducing diseases.

Our analysis adds to the existing evidence by providing practical recommendations for using AI, specifically in health promotion and disease reduction strategies. The key identified strategies include conducting further investigations, taking into account the needs of end users, improving the technical aspect of the technology, and allocating resources. It is recommended that mitigating the weaknesses of AI in health promotion, such as technical limitations and data quality concerns, includes structured interdisciplinary collaboration [121]. Engaging health care professionals ensures that AI applications align with clinical workflows and patient needs. Data scientists and AI engineers contribute by refining algorithmic accuracy and robustness through improved data preprocessing and validation techniques. Ethicists and policymakers play a crucial role in establishing governance frameworks that enhance data transparency and security while addressing potential biases. Moreover, integrating public health experts and community representatives into AI development can improve dataset diversity, ensuring AI systems are trained on representative populations. By fostering structured interdisciplinary collaboration, AI tools can become more reliable, equitable, and aligned with real-world health care challenges [122,123]. In alignment with these recommendations, several guidelines and road maps have been published. Although these focus on AI implementation in health care systems, they can serve as a guiding light for leveraging AI to promote health and mitigate diseases. These tool kits offer best practices for the implementation of AI in health care, address key challenges in AI adoption, offer recommendations for optimizing AI deployment in health systems [124], and recommend practical guidance for AI governance to help maximize the benefits of AI while minimizing its foreseeable risks [125].

Potential Impact and Future Directions

To harness the potential of AI in health promotion and disease reduction, recent advancements in AI technologies should be acknowledged. Innovative AI tools, such as AI-based virtual health assistants (eg, Lucy [126], Paola [65]), interactive chatbots (Woebot [127], Replika [128]), digital health coaching systems (Lark [129], Ginger [130]), and diagnostic apps (Ada Health [131], Babylon Health [132]), are emerging as promising solutions for enhancing personalized health advice, supporting behavior change, and improving chronic disease management. Highlighting such advancements can inspire future research to explore their broader applicability in diverse health care settings. Future studies should focus on assessing the effectiveness, scalability, and ethical implications of these technologies in real-world contexts.

First, we believe that to facilitate using AI in the field of health promotion and disease reduction, it is essential to conduct real-world studies to assess the effectiveness and long-term sustainability of AI interventions across diverse populations, given the observed benefits in areas like smoking cessation, physical activity, and dietary improvement. Second, addressing user engagement and trust issues is critical; although AI interventions have demonstrated high usability, concerns around trust and emotional connection highlight the need for transparent, user-centered design to enhance acceptance and lead to consistent engagement. Finally, economic evaluations are crucial to determine the cost-effectiveness and scalability of AI applications in health care systems, as these interventions have shown promise in reducing health care costs and the workload of health professionals. Future studies should incorporate structured economic evaluations, such as cost-benefit and cost-effectiveness analyses, to assess the financial viability of AI-driven health interventions and guide resource allocation and inform policy decisions on sustainable AI adoption.

In addition to these recommendations, the seamless integration of AI technologies into existing health care workflows is essential to maximize their benefits. Policymakers and health care stakeholders should prioritize regulatory frameworks that support the ethical and equitable adoption of AI while ensuring alignment with clinical guidelines and public health priorities. It is worth mentioning that effective AI integration requires collaboration among policymakers, health care providers, technology developers, and public health officials to address ethical, technical, and infrastructural challenges. AI initiatives should be designed with interoperability in mind to facilitate adoption in diverse health care settings. These findings align with global digital health strategies, such as those outlined by the World Health Organization, which emphasize the role of AI in strengthening health care systems, improving accessibility, and supporting data-driven decision-making. However, future research should further explore policy-driven approaches to ensure AI adoption is both sustainable and ethically sound across different health care contexts.

Limitations of the Study Design

The main limitation of this review is its design. Contrary to systematic reviews, but consistent with protocols for expedited reviews [37], we narrowed the scope of the search to align with the study’s objectives by using limited databases and imposing a publication date restriction. These methodological choices allowed for appropriate and structured study selection, data extraction, and critical appraisal in a timely manner. However, it is important to note that this rapid review has limitations.

To ensure that stakeholders and decision-makers receive the most current evidence, the search was restricted to studies published in the last 5 years and limited to Organisation for Economic Co-operation and Development countries. Although this approach enhances relevance to contemporary AI applications, it may affect the generalizability of findings, particularly to low- and middle-income countries, where health care infrastructure, digital literacy, and AI accessibility differ significantly. Additionally, several included studies had small sample sizes, which may limit the robustness and generalizability of the findings. The reliance on small cohorts can introduce selection bias and reduce the statistical power needed to validate AI-driven interventions in diverse populations.

Moreover, the review may be subject to publication bias, as studies reporting positive outcomes are more likely to be published than those with null or negative findings. Although we acknowledged this risk, we did not incorporate gray literature or preprints, which could have improved the comprehensiveness of our evidence base. Therefore, future research should expand its scope by incorporating a broader range of countries, more diverse databases, and larger, more representative samples. Future reviews should also integrate mitigation strategies, such as the inclusion of gray literature, to enhance methodological rigor and reduce potential publication bias.

Furthermore, no formal quality appraisal or risk of bias assessment was conducted for the included studies. Although this is consistent with common practices in rapid reviews, it limits the ability to assess the strength and certainty of the synthesized evidence. Future studies should incorporate standardized quality assessment tools to improve the credibility of findings.

Lastly, while our review focused on the application of AI in health promotion and disease reduction, we acknowledge that a detailed technical analysis of AI models, including their accuracy, efficiency, and scalability, was beyond the scope of this study. Future research should explore these aspects, particularly by integrating causal inference methods such as propensity score matching, instrumental variable analysis, and difference-in-differences to establish the direct impact of AI interventions on health outcomes. Such approaches would provide stronger evidence on whether AI-based health interventions yield superior results compared to traditional methods. To bridge gaps in long-term outcomes and strengthen the evidence base for AI-driven health promotion interventions, future research should use robust study designs such as longitudinal studies and randomized controlled trials. These methodologies would provide deeper insights into the sustained impact of AI interventions over time and improve causal inferences regarding their effectiveness.

Conclusion

The integration of AI and related technologies holds considerable potential to enhance health promotion efforts and prevent chronic conditions through personalized, scalable, and interactive interventions. This review synthesized existing evidence and practical applications demonstrating the effectiveness of AI in supporting physical activity, dietary improvements, smoking cessation, and mental well-being across diverse populations. Importantly, the findings of this review highlight several critical priorities for future policy and implementation. First, personalization and real-time engagement are central to the success of AI-based interventions, though maintaining long-term user interaction and trust remains a key challenge. Second, while AI tools offer significant scalability and remote accessibility, their effectiveness in underserved communities is contingent upon addressing infrastructural and digital literacy barriers. Third, ethical concerns—including algorithmic bias, data privacy, and informed consent—must be proactively addressed through inclusive design, regulatory oversight, and transparent governance mechanisms. Additionally, the lack of long-term impact assessments and comparative effectiveness studies underscores the need for robust real-world research and economic evaluations to guide sustainable adoption. The SWOT analysis presented in this review offers a strategic framework to inform implementation pathways. By aligning AI deployment with health care system priorities and involving multidisciplinary stakeholders, including health care providers, public health professionals, developers, and policymakers, AI can be more effectively integrated into clinical and community-based health promotion strategies. This synthesis thus provides a foundation for designing equitable, efficient, and ethically sound AI-driven health interventions and supports decision-makers in translating AI innovations into improved population health outcomes.