In conducting a systematic review between November and December 2023, we adhered to the PRISMA (Preferred Reporting Items for Systematic Reviews and Meta-Analyses) guidelines [19] (Multimedia Appendix 1 and Checklist 1).

Comprehensive searches were performed across databases including PubMed, Scopus, and Web of Science, using Health Science Descriptors and Medical Subject Headings descriptors pertinent to AI in health care and associated skills. The search strategy was refined using Boolean and truncation operators. The search queries included combinations such as (“Artificial Intelligence” OR “AI”) AND “Healthcare Professionals” AND (“Skills” OR “Competencies” OR “Education”), (“Artificial Intelligence” OR “AI”) AND (“data analysis” OR “ethical considerations”).

We established specific selection criteria, prioritizing peer-reviewed original articles, review articles, editorials, and commentaries that directly explored the required skills for health care professionals to integrate AI into their practice. Regarding the inclusion criteria, studies were considered if they: were published in English or Spanish, taking advantage of the linguistic accessibility for the research team in order to reflect the possible specific characteristics of the Spanish-speaking community; published from 2018 onward, to ensure the relevance and timeliness of the research given the rapid advancements in AI technologies in recent years; focused on the necessary skills for the effective use of AI by health care professionals, encompassing both technical and management competencies; and included aspects related to training in digital health, highlighting the importance of specific training in the use of emerging technologies.

Conversely, studies were excluded if they did not meet these criteria: studies written in languages other than English or Spanish, as they could not be accurately analyzed by the research team; studies published before 2018, to focus on the most recent trends in AI in health care; studies not specifically related to the skills or needs of health care professionals for the use of AI tools, excluding research that did not directly address this focus; and studies that did not refer to specific skills or training in digital health, discarding those that did not directly contribute to understanding the competencies necessary to integrate AI into health care practice.

This selection methodology was designed to identify studies that provided significant evidence on the key competencies health care professionals need to develop for effectively integrating AI into their clinical practice, thus, ensuring that the systematic review focused on research offering practical and applicable insights.

The search strategy was developed iteratively to optimize the retrieval of relevant studies. An initial search used the aforementioned databases and descriptors. Titles and abstracts underwent rigorous review for relevance by 2 independent researchers (JG-G and CLS-B), with duplicates removed and articles not meeting inclusion criteria or fitting exclusion criteria discarded. Mendeley (Elsevier Ltd) served as the reference management tool, not involved in the data extraction process.

Subsequently, full-text articles were retrieved for an in-depth content review based on the inclusion criteria specified earlier. However, 3 articles were not retrievable despite repeated efforts. Two articles were inaccessible due to subscription restrictions, and attempts to obtain them via interlibrary loans were unsuccessful. The third article had a broken link, and the authors were unresponsive after multiple contact attempts. These articles have been documented in the “reports not retrieved” section of the PRISMA flow diagram in the Results section.

To assure methodological integrity, discrepancies were resolved through consultation with a third researcher (JLS), reaching consensus on study admissibility. The screening process was manual, without the use of automation tools, and each study was carefully evaluated. Mendeley was used again for reference management, without affecting the data extraction process. The PRISMA flow diagram is provided in the Results section.

Data were extracted and synthesized from eligible studies, encompassing details such as authors, publication year, study design, identified skills, and methodological quality assessment. Tables served as the primary method for tabulating and visually presenting the results and their synthesis.

The data search spanned all pertinent dimensions of these outcomes, including variables relevant to the analyzed studies. These variables covered aspects such as humanization, social skills, and participants’ AI usage experience, offering a comprehensive perspective on the competencies necessary for integrating AI into health care practices.

The GRADE (Grading of Recommendations Assessment, Development, and Evaluation) framework [20] was applied to assess the evidence quality of the included works, considering the quality of studies, result consistency, imprecision, potential bias, and other pertinent factors. Mixed methods or qualitative studies were appraised using the mixed methods appraisal tool [21].

Bias in the examined works was evaluated following Cochrane’s domain-based recommendations [22], considering five types of bias, each with its domains. Bias risk was independently assessed by at least 2 reviewers, with discrepancies resolved via discussion or consultation with a third reviewer when necessary.

The research question guided the synthesis groupings, focusing the analysis on the skills and competencies necessary for the effective implementation of AI technologies by health care professionals. No standardization metric was applied. The synthesis method involved extracting relevant sections of the studies related to the identified skills and competencies. Overall, the risk of bias assessment of the studies showed no critical results. Consequently, the included studies were synthesized with equal weight.

The categorization of skills and competencies was based on a qualitative synthesis approach, where 2 independent reviewers extracted and coded data from each study. The categories were developed iteratively by grouping competencies that were conceptually aligned. The final domains were determined based on the frequency with which competencies appeared across the studies, with higher-frequency competencies categorized as key domains. For example, “AI fundamentals” and “ethical and legal considerations” were identified as critical due to their frequent mention in the selected studies, whereas others such as “data governance” and “programming” appeared less often but were still considered relevant.

Study design was not restricted, so a meta-analysis was not performed due to the heterogeneity of the studies and the differing amounts and qualities of information reported on skills and competencies. The reported effectiveness of the competencies was synthesized based on each study’s findings. The identified competencies are presented in the results section. Tables and figures aggregate information about this study’s characteristics and focal areas of this review (skills and competencies).

This review aimed to identify and delineate the skills and competencies essential for health care professionals to use AI in their clinical practices effectively. The objective encompassed various outcome domains, not limited to interpreting results from machine learning models, managing AI biases, ethical considerations in AI-assisted decision-making, and the technical skills required for effective AI tool use.

The initial database searches yielded a total of 2457 articles (Figure 1). After deleting duplicates and screening titles and abstracts, 37 articles were selected for full-text review. Out of these, only 7 met all the inclusion criteria for this systematic review [23-29]. Each selected article specifically concentrated on the competencies essential for the integration of AI into routine clinical practice. A substantial number of the excluded articles (n=28) made reference to, but did not directly engage with, the competencies in question. Table S1 in Multimedia Appendix 1 delineates the characteristics of these studies and the principal competencies identified therein.

The review identified a diverse range of skills and competencies necessary for the effective implementation of AI in health care. These competencies were categorized into several key areas based on their frequency of appearance in the selected studies. The identified skills, definitions, and the number of studies that reported each are summarized in Table S2 in Multimedia Appendix 1.

The analysis of the selected studies highlights several key competencies deemed essential for the effective integration of AI into routine clinical practice.

AI fundamentals were identified as an essential competency by 86% of the studies [23-27,29]. This indicates a strong consensus on the importance of foundational AI knowledge for health care professionals. The studies emphasize that a solid understanding of AI principles is crucial for effectively implementing and using AI technologies in clinical settings. This foundational knowledge forms the bedrock upon which more advanced competencies are built, ensuring that professionals can confidently engage with AI tools and integrate them into their practice.

Ethical and legal considerations were emphasized by 71% of the studies [23-26,29]. The studies underscored the importance of having a solid knowledge base in various aspects of ethics, including patient privacy, data security, biases in algorithms, and transparency and explainability.

Data analysis and management skills were highlighted by 43% of the studies [24,26,28]. These studies emphasize several secondary skills crucial for effective data handling. For instance, McCoy et al [26] stress the importance of data acquisition, cleaning, and visualization as foundational steps in preparing data for AI applications. Singh et al [24] underline the need for robust data management practices to ensure the integrity and reliability of AI outputs. Wiljer and Hakim [28] highlight the necessity of developing capabilities in data governance, which includes the secure storage and regulatory compliance of health care data.

Communication and teamwork were identified as important competencies by 43% of the studies [23,25,27]. These studies underscore the critical need for effective communication skills to convey complex AI-related information to both colleagues and patients. Çalışkan et al [23] emphasize the role of interdisciplinary teamwork in the development and implementation of AI applications, highlighting the necessity for seamless collaboration between health care professionals and AI experts. Liaw et al [25] point out the importance of clear and empathetic communication with patients regarding the use of AI in their care, ensuring transparency and maintaining trust. Sujan et al [27] stress the need for health care teams to work cohesively to monitor and supervise AI systems, ensuring their safe and ethical use.

Evaluation of AI tools was mentioned by 43% of the studies [25,26,29]. These studies highlight several essential secondary skills necessary for the rigorous assessment of AI technologies. Liaw et al [25] emphasize the importance of understanding evidence-based evaluation methods to critically assess the performance and utility of AI tools in clinical practice. McCoy et al [26] discuss the need for health care professionals to be proficient in performing critical evaluations, including assessing the accuracy, reliability, and validity of AI algorithms. Sapci and Sapci [29] stress the significance of ongoing evaluation and monitoring of AI tools to ensure they consistently meet clinical standards and enhance patient outcomes.

Some investigations highlight more technical abilities, such as programming [26], or a deeper mathematical acumen [24,28]. These skills, while crucial, were not as frequently cited as the previously mentioned competencies. Conversely, a capability deemed highly significant across the literature is the aptitude for evaluating AI tools to ascertain their quality and rationalize their application, in addition to scrutinizing potential biases and limitations [24,25].

Figure 2 presents a mosaic plot displaying the distribution of identified competencies based on the percentage of studies that reported each skill. The size of each tile corresponds to the proportion of studies mentioning that competency, providing a visual representation of the relative importance of each skill across the reviewed literature.

This review underscores the critical need for health care professionals to acquire competencies related to the effective use of AI in clinical practice. Five key areas of competence were identified as essential for AI integration in clinical practice: AI fundamentals, ethical and legal considerations, data analysis and management, communication and teamwork, and the evaluation of AI tools. These competencies are essential to ensure the safe and reliable integration of AI technologies into routine health care workflows. Furthermore, ethical and legal considerations, such as data security and transparency, are crucial for building trust in AI-driven decisions (Table S2 in Multimedia Appendix 1).

In line with our findings, recent studies have emphasized the growing need to define and standardize AI competencies within clinical settings. A 2022 exploratory review [30] highlighted this gap, calling for more structured educational programs to support the integration of AI into health care practice. Additionally, a survey among nursing staff [31] indicated that most professionals had acquired their AI skills independently, further reinforcing the need for formalized training curricula.

Recent data from Rock Health, a venture capital firm specializing in digital health, have underscored an exponential increase in investments directed toward digital health enterprises and related technologies [42]. This trend distinctly signals an evolving health care landscape, increasingly reliant on novel technologies, thereby accentuating the imperative for health care professionals to proficiently integrate such advancements into their clinical practices [43].

Despite widespread agreement on the necessity for comprehensive AI training from the outset of medical education, there is a lack of consensus on the specific content and approach of such training [15]. The discussion around this issue is abundant, yet concrete resolutions are rare. The findings of this review aim to provide an approach to the possible competencies required, offering a structured framework for developing comprehensive AI training programs for health care professionals.

Integrating AI education into health care curricula presents several challenges due to the variability and lack of standardization of required competencies. To address this need, we propose general guidelines for the development of essential AI competencies for health care professionals. These guidelines provide a reference framework that educators and course coordinators can use to design training programs, assess student progress, and establish performance criteria. For example, health care professionals should acquire the ability to assess the quality of algorithms and their interpretations, as well as identify and mitigate potential biases. This approach facilitates the integration of AI into clinical practice and provides a clear guide for curriculum development and performance evaluation.

Augmenting the education and training of health care professionals is posited to elevate their confidence in using these tools. Although concerns persist regarding AI’s potential to supplant human roles, a more discerning view proposes that AI will primarily alleviate the burden of mundane tasks. This reallocation of time and resources is anticipated to enhance patient interactions and elevate the quality of health care services provided [39].

Several authors argue that the existing academic infrastructure is ill-equipped to incorporate AI education, citing time constraints and a lack of teaching expertise as significant obstacles. An alternative proposed involves the use of specific AI tools not only for clinical applications but also to elucidate the underlying algorithms, focusing on their practical use and ethical implications [44].

A solution for integrating AI education into health care curricula, addressing the shortage of instructors with expertise in clinical AI applications, involves leveraging established training programs from other institutions [45]. Programs by Stanford University [46] and Harvard University [47] serve as examples, providing access to high-quality educational content. These programs offer a comprehensive curriculum that covers essential AI concepts, practical applications, and ethical considerations, enabling health care professionals to gain a deep understanding of AI technologies.

The integration of AI into clinical practice is expected to augment, not replace, the roles of health care professionals. It calls for a workforce proficient in digital health and communication, capable of leveraging AI’s benefits while recognizing its limitations and ethical considerations [48]. This paradigm shift offers an opportunity to enhance patient care, delegating computational tasks to AI and focusing on the human aspects of health care delivery [37].

This systematic review encounters several challenges, primarily due to the limited availability of literature specifically addressing the competencies required for integrating AI into health care practice. The scarcity of targeted studies can be attributed to the nascent and rapidly evolving nature of AI applications in health care.

To mitigate this issue, the search criteria were broadened to include general terms related to AI in health care and competencies required by health care professionals. While necessary, this broadening may have introduced studies that do not exclusively focus on AI competencies, potentially affecting the homogeneity of the findings. The review also faced potential language bias, as it primarily focused on literature in English and Spanish. Pertinent studies in other languages might have been excluded. Despite rigorous and independent review processes, the selection could still be influenced by subjective interpretation of the inclusion and exclusion criteria.

Limiting the review to studies published after 2018 aimed to capture the most recent advancements, but this restriction might omit emerging research. Despite these efforts, there remains a clear knowledge gap in the specific skills required for effective AI use in health care. The literature predominantly addresses AI applications and benefits, but lacks detailed research on the precise skills health care professionals need. Current research is often fragmented and varies significantly in scope and depth. Few studies offer comprehensive models or curricula for AI competency training in health care. This inconsistency highlights the need for more robust research to identify essential AI skills and explore effective methods for integrating these skills into health care education and practice.

This systematic review has identified essential competencies for health care professionals to effectively integrate AI into clinical practice. The analysis reveals a consensus on the importance of five key areas: AI fundamentals, ethical and legal considerations, data analysis and management, communication and teamwork, and evaluation of AI tools. These competencies are crucial for leveraging AI technologies to enhance patient care and health care delivery. A consensus within the scholarly discourse suggests the necessity for health care professionals to attain proficiency in these domains to ensure the judicious application of AI tools, thereby accruing benefits for both patients and the health care ecosystem.

AI fundamentals form the backbone of necessary knowledge, enabling health care professionals to understand and use AI technologies effectively. Ethical and legal considerations ensure that AI applications adhere to patient privacy, data security, and transparency standards, maintaining trust and compliance within health care settings. Data analysis and management skills are vital for handling large datasets, ensuring accurate AI outputs, and supporting informed clinical decisions.

Communication and teamwork are also critical, facilitating the clear conveyance of AI-related information among health care professionals and to patients, thereby promoting transparency and trust. The ability to evaluate AI tools is essential for assessing the performance and reliability of AI technologies, ensuring they meet clinical standards and deliver safe, effective patient care.

Augmenting the education and training of health care professionals is posited to elevate their confidence in using these tools. Although concerns persist regarding AI’s potential to supplant human roles, a more discerning view proposes that AI will primarily alleviate the burden of mundane tasks. This reallocation of time and resources is anticipated to enhance patient interactions and elevate the quality of health care services provided.

In this context, the importance of communication skills becomes increasingly paramount. The introduction of AI tools is expected to afford health care professionals additional time per patient encounter, potentially heightening patient satisfaction and care quality.

The ambition extends beyond merely acquiring proficiency in disciplines ancillary to traditional health care paradigms. Considering the already intricate and comprehensive nature of health care education, particularly in medicine, the emphasis is placed on fostering an in-depth comprehension of AI’s functionalities, inherent biases, pragmatic utility, and cost-effectiveness compared to abstaining from AI applications.

The integration of AI into health care is indispensable for advancing patient care but requires a concerted effort to develop and standardize competencies among health care professionals. Regulatory oversight and enhanced educational frameworks are essential for overcoming existing barriers and leveraging AI’s full potential in clinical settings.

Despite the progress highlighted in this review, significant gaps remain in the literature, particularly concerning the specific educational frameworks and training programs needed to develop these competencies. Most existing research focuses on the potential applications and benefits of AI, with less emphasis on the precise skills required for effective implementation.

Future research should prioritize the development and validation of standardized AI competency frameworks tailored for health care professionals. These frameworks should cover technical skills, ethical and legal aspects, and data management practices. Collaborative efforts between academic institutions, health care organizations, and AI experts can create comprehensive training programs to address these competencies.

Additionally, longitudinal studies are necessary to evaluate the long-term effectiveness of AI training programs. Research should explore how health care professionals apply their AI training in clinical settings, assessing the impact on patient outcomes, clinical decision-making, and health care efficiency.