We conducted a systematic search of the literature in accordance with the PRISMA (Preferred Reporting Items for Systematic Reviews and Meta-Analyses) 2020 guidelines (Checklist 1). Five electronic databases—PubMed or MEDLINE, Web of Science, Scopus, Google Scholar, and the Saudi Digital Library—were searched to identify studies evaluating the impact of AI on health care services in Saudi Arabia. The search combined controlled vocabulary terms (eg, Medical Subject Headings) and free-text keywords related to AI and health care. The following concepts were applied: “artificial intelligence” OR “machine learning” OR “deep learning” AND “healthcare” OR “health services” OR “healthcare quality” OR “diagnostic accuracy” AND “Saudi Arabia” OR “Saudi healthcare.” Additional terms were included to capture provider-level perspectives, such as “healthcare providers” OR “clinical decision support.”

The search covered the period from January 2014 to April 2024 to ensure contemporary relevance, reflecting the accelerated adoption of AI in Saudi health care during the past decade. No language restrictions were applied at the search stage; however, only English language full texts were included at the screening stage. Boolean operators (AND and OR) and database-specific filters were used to refine results. The reference lists of all included studies were also hand-searched to identify additional relevant publications.

This study is based on original research and academic papers from journals. The Mendeley References Manager software (Elsevier) identified all the records from the database to display the collected data [13]. The data were then entered into a Microsoft Excel worksheet to locate all the studies. After that, all titles were screened on the basis of the inclusion and exclusion criteria. Abstracts of the articles were analyzed in depth to ensure the quality and relevance of the academic literature included in the review process. Full-text screening was then done, and all these steps were presented on the PRISMA flow diagram. The systematic review followed the widely accepted PRISMA 2020 guidelines, which serve as a comprehensive flow diagram to ensure transparency and reproducibility [14]. The PRISMA flow diagram visually represents the study selection process from initial identification to the final inclusion of studies in the review.

To ensure the quality of the review, each included research paper was evaluated carefully at a later stage through the Joanna Briggs Institute (JBI) critical appraisal checklist [15]. The JBI critical appraisal checklist enhances systematic reviews by ensuring methodological rigor, assessing bias, and improving the reliability of included studies. It supports transparent and reproducible research, strengthening evidence-based health care decisions. The methodological quality was assessed in the studies chosen using the JBI methodology’s tailored criteria for different kinds of research. We performed the critical appraisal of eligible analytical cross-sectional studies for 8 studies, the critical appraisal of eligible cohort study research for 3 studies, and the critical appraisal of quasiexperimental studies for 1 study.

At the beginning of the review process, 874 records were identified across 3 databases: 295 from Google Scholar, 293 from PubMed, and 286 from the Saudi Digital Library (Table 1). Of the total, 26 duplicates were removed, leaving 848 unique records. Following title screening, 748 records were excluded for irrelevance and 99 abstracts were reviewed. Of these, 22 were excluded due to ineligible study design and 56 because they were not conducted in Saudi Arabia; 22 full texts were assessed, of which 9 were excluded (6 for ineligible populations and 3 for design). A final total of 12 studies were included. The PRISMA flow diagram is provided in Figure 1.

The included studies addressed 4 major themes: diagnostic accuracy (n=3), patient management (n=3), operational efficiency (n=3), and quantified impact (n=3; Table 2). These themes capture the breadth of AI applications in Saudi health care and highlight the diverse ways AI has been studied, from early diagnostic models to workforce efficiency and provider perceptions (Multimedia Appendices 1 and 2).

Three studies demonstrated that AI can improve diagnostic precision in Saudi health care contexts by leveraging clinical, imaging, and laboratory data. In the context of the COVID-19 pandemic, Aljameel et al [16] applied a random forest classifier to records from King Fahad University Hospital and achieved 95% accuracy and an area under the curve (AUC) of 0.99, suggesting that AI could serve as an effective early-warning tool for predicting disease severity and outcomes. This has direct implications for triaging patients in high-demand hospital settings. In oncology, Ibrahim et al [17] developed the AMAN detection system for interpreting equivocal mammograms, reporting 87% image classification accuracy, a 95% AUC, and 92% balanced accuracy, thereby confirming the feasibility of cost-effective deep learning–based systems to support breast cancer screening. In neurological care, a study tested 4 machine learning algorithms on hospital-based clinical and laboratory records, with support vector machines achieving 95.6% accuracy and an area under the receiver operating characteristic curve of 0.97 [22]. These results highlight the potential of AI to detect Alzheimer disease at earlier stages, where interventions may have a greater impact. Collectively, these studies show how AI can be integrated across diverse diagnostic pathways in Saudi Arabia, from infectious disease to oncology and neurology.

Three studies assessed how AI supports personalized care and treatment planning. Olatunji et al analyzed data from 569 patients with multiple sclerosis using explainable AI, a method that not only produced robust predictions but also provided insight into the decision-making process of the algorithms. This interpretability allowed clinicians to better understand the rationale for predictions, strengthening trust and adoption in clinical settings. Alahmadi et al [23] developed a gradient boosting model using records of 622 patients with COVID-19, achieving 96.3% accuracy and identifying 23 predictive features, including comorbidities and beta-blocker use, which may reduce disease severity. This work underscores the value of AI for risk stratification and treatment prioritization during public health emergencies. Al Yousef et al [12] applied data mining techniques to electronic health records from 5 National Guard Health Affairs hospitals to predict diabetes onset. While the models performed acceptably, missing data limited accuracy; however, the study demonstrated the feasibility of using routinely collected hospital data for early disease prediction in Saudi Arabia. These examples illustrate the expanding role of AI in chronic disease management and highlight its potential for precision medicine approaches.

Three studies investigated AI applications for workflow optimization and organizational performance. Al Habib and Alharthi [9] used machine learning to predict physician noncompliance with electronic health record documentation. The random forest model achieved an AUC of 0.891, an F₁-score of 0.831, and a recall of 0.831, demonstrating strong predictive performance and offering a practical solution for hospitals to proactively address compliance issues. Seliaman evaluated physician and pharmacist perceptions of clinical decision support systems (BESTCare 2.0), which are widely implemented in the National Guard hospitals. Findings indicated that usability, availability of patient information, and accessibility were critical factors influencing acceptance, highlighting both the potential of clinical decision support systems and the challenges in integrating them seamlessly into clinical workflows. In dentistry, Alharbi and Almutiq [6] explored AI models for predicting dental implant cases, showing that AI could improve treatment planning and reduce complications, thereby streamlining dental services. Together, these studies suggest that AI has important applications not only in clinical care but also in enhancing hospital efficiency and administrative oversight.

Three studies measured the broader impact of AI adoption in health care delivery. Alghamdi et al [11] reported that 47% of health care providers surveyed had used telehealth applications, but barriers such as poor internet connectivity, lack of time, and insufficient training limited widespread use. Serbaya et al [20] conducted a survey of health care workers in Jeddah and found that while 75% understood AI principles, only 50% were willing to use AI in medical decision-making. This gap highlights cultural and trust challenges that could slow AI adoption, even in settings where technical capability exists. Aldossari [21] showed that younger physicians, particularly those aged <30 years, expressed higher satisfaction with telehealth and demonstrated greater openness to expanded AI use in their practice. These findings emphasize the need for targeted strategies to improve provider engagement, training, and trust in AI tools.

Critical appraisal indicated moderate to high quality across most included studies. Cross-sectional studies scored between 50% and 100% across appraisal items, while cohort and quasiexperimental designs achieved scores ranging from 27% to 72%. Habib and Alharthi [9] achieved the highest appraisal score (100%), reflecting rigorous methodology, comprehensive outcome measures, and strong predictive performance. Although variability in quality was observed, particularly among quasiexperimental studies, the majority demonstrated methodological soundness and provided valuable insights into the role of AI in Saudi health care (Tables 3 and 4). Only one quasiexperimental study (Ibrahim et al [17]) met the inclusion criteria. The study satisfied 4 of 9 quality appraisal items (44%), with clear reporting for participant selection and outcome measurement but limited information regarding confounding control and intervention description.

This systematic review identified 12 eligible studies that examined the impact of AI on health care services in Saudi Arabia. The main findings show that AI applications have primarily focused on 4 domains: diagnostic accuracy, patient management, operational efficiency, and quantified impact. Across these studies, AI consistently demonstrated potential to improve diagnostic precision, support clinical decision-making, streamline health care workflows, and enhance cost-effectiveness. The integration of AI into health care represents a transformative development with profound implications worldwide, including in Saudi Arabia. The establishment of the Saudi Data and Artificial Intelligence Authority in August 2019 by royal decree marked a significant milestone in advancing the nation’s digital transformation under [2,8]. AI’s capacity to enhance diagnostic accuracy, streamline health care processes, and reduce operational costs is well documented. Johnson et al [10] examined the impact of AI on the health care sector in the United States, highlighting several advantages that align with findings observed in Saudi Arabia.

AI-driven models have demonstrated significant advancements in medical diagnostics, enabling more precise disease identification and improving patient outcomes. For instance, Aljameel et al [16] developed a machine learning model for predicting COVID-19 outcomes, achieving 95% accuracy with an AUC of 0.99, thereby demonstrating AI’s potential in pandemic response. Similarly, Ibrahim et al [17] introduced the AMAN system for breast cancer detection, reporting an 87% accuracy in mammography and a 95% AUC, reinforcing AI’s role in oncological diagnostics. These findings align with global efforts to enhance diagnostic precision through AI-driven methodologies.

AI has been instrumental in advancing patient management by facilitating early disease detection and optimizing treatment strategies. Olatunji et al [18] developed a machine learning model for early multiple sclerosis diagnosis, allowing for timely intervention and personalized treatment, which is critical for managing chronic conditions. Alghamdi et al [11] examined telehealth adoption among health care providers, revealing that 47% of surveyed professionals used telehealth, although barriers such as limited time availability and limited connectivity impeded broader implementation [26-28]. These insights underscore AI’s role in enhancing health care accessibility and personalized care delivery.

Beyond clinical applications, AI improves operational workflows and compliance monitoring. Al Habib and Alharthi [9] applied machine learning techniques to predict electronic health record documentation compliance, achieving an AUC score of 0.891, thereby assisting in identifying noncompliant physicians and supporting regulatory adherence. Furthermore, Alharbi and Almutiq explored AI applications in predicting dental implant cases, demonstrating AI’s ability to enhance patient outcomes and reduce complications associated with dental procedures [3]. These advancements underscore AI’s role in improving the efficiency of health care systems.

AI integration extends to health care data security, cost reduction, and competitive advantages. Almalawi et al [7] emphasized the importance of secure health care data management, highlighting AI’s potential in mitigating cybersecurity risks while improving efficiency in health care administration. Similarly, Al Yousef et al [12] developed predictive models for early diabetes diagnosis, aiding in proactive disease prevention and effective health monitoring despite persistent technical challenges. AI-driven predictive analytics can significantly reduce health care costs and contribute to data-driven decision-making models.

Despite these advancements, notable differences exist between Saudi Arabia and the United States regarding data privacy and security concerns. While both countries acknowledge the importance of data protection, their regulatory frameworks diverge significantly. Saudi Arabia aligns AI integration with National Data Governance Regulations, fostering a centralized approach to data protection. In contrast, the United States operates within a multifaceted regulatory landscape that includes the Health Insurance Portability and Accountability Act (HIPAA) alongside various federal and state privacy laws, resulting in a more fragmented approach to data governance [29].

Resistance to AI adoption among health care providers has also been observed across both regions. In Saudi Arabia, skepticism is often linked to concerns about the potential erosion of the human element in patient care and the technological adaptations required for AI integration [8]. Conversely, in the United States, reluctance is more commonly associated with apprehensions regarding job displacement and challenges in integrating AI with existing health care IT infrastructure [10].

Several factors, including professional roles and cultural attitudes, influence AI adoption. A study by Serbaya et al [20] found that pharmacists exhibited a higher level of trust in AI compared to physicians and nurses, a trend also observed in European health care settings. Pharmacists’ familiarity with medication management technologies and their perception of AI-assisted decision-making as objective may contribute to this pattern. Additionally, gender disparities in AI confidence levels have been documented, with male health care professionals displaying greater confidence in AI applications than their female counterparts. Findings from previous studies attribute this discrepancy to greater exposure to technology among male professionals, job roles requiring technological engagement, and societal norms influencing interactions with emerging technologies [13,30,31]. Addressing these disparities through targeted educational programs and inclusive AI development strategies is crucial for equitable technology adoption.

The following are actionable recommendations for stakeholders:

The following are recommendations to improve infrastructure and training:

The following are recommendations to foster collaboration and innovation:

The following are regulatory suggestions for AI implementation:

The following are challenges with regard to data quality and availability:

The following are challenges with regard to regulations and ethical guidelines:

The following are challenges faced in providing training for medical personnel:

Integrating AI into health care policy and enhancing the continuous use of AI health devices requires a multifaceted approach. By defining clear objectives, involving stakeholders, and focusing on user-friendly design and constant monitoring, we can ensure that AI technologies are effectively used to improve health care outcomes. Addressing challenges such as data privacy, infrastructure investments, and bias mitigation will be crucial for successful implementation.

AI can potentially revolutionize health care in Saudi Arabia by predicting patient outcomes; personalizing medicine; streamlining administrative tasks; and improving accuracy, efficiency, and cost-effectiveness. However, it is critical to also focus on proper training for medical personnel while maintaining data privacy and addressing biases in AI algorithms. The Saudi Arabian health care sector can benefit from AI in many ways, including predicting disease outbreaks, enhancing patient admissions processes, and supporting ongoing education for medical professionals. Furthermore, deploying telehealth services and other digital health strategies highlights AI’s vital role in the region’s health care future. To fully realize AI’s potential in health care, it is essential to tackle the associated challenges and risks. With careful implementation and training, AI can lead to more accurate and efficient health care services, improving patient outcomes and reducing costs.