Artificial intelligence in disease diagnostics: a comprehensive narrative review of current advances, applications, and future challenges in healthcare.

AI in healthcare represents a convergence of computer systems designed to emulate human cognitive functions such as learning, reasoning, and problem-solving [ 10 ] . This section delves into various AI technologies and their applications in the healthcare sector. ML, a core component of AI, utilizes statistical methods to analyze and interpret large datasets, enabling groundbreaking discoveries in healthcare [ 11 ] . ML models are broadly categorized into three types: Supervised learning: These models are trained on labeled datasets, where the input data is paired with the correct output. Supervised learning is widely used for tasks such as classification (e.g. diagnosing diseases) and regression (e.g. predicting patient outcomes). For example, supervised learning algorithms have been employed to predict the risk of heart disease based on patient demographics and medical history [ 12 ] . Unsupervised learning: Unlike supervised learning, unsupervised learning models work with unlabeled data, identifying patterns and groupings within the dataset. These models are particularly useful for clustering patients with similar characteristics or detecting anomalies in medical data. For instance, unsupervised learning has been applied to group patients with similar genetic profiles for personalized treatment planning [ 12 ] . Reinforcement learning: This approach involves training models to make sequential decisions based on rewards and penalties. Reinforcement learning has shown promise in optimizing treatment strategies, such as determining the optimal dosage of chemotherapy for cancer patients [ 12 ] . Supervised learning: These models are trained on labeled datasets, where the input data is paired with the correct output. Supervised learning is widely used for tasks such as classification (e.g. diagnosing diseases) and regression (e.g. predicting patient outcomes). For example, supervised learning algorithms have been employed to predict the risk of heart disease based on patient demographics and medical history [ 12 ] . Unsupervised learning: Unlike supervised learning, unsupervised learning models work with unlabeled data, identifying patterns and groupings within the dataset. These models are particularly useful for clustering patients with similar characteristics or detecting anomalies in medical data. For instance, unsupervised learning has been applied to group patients with similar genetic profiles for personalized treatment planning [ 12 ] . Reinforcement learning: This approach involves training models to make sequential decisions based on rewards and penalties. Reinforcement learning has shown promise in optimizing treatment strategies, such as determining the optimal dosage of chemotherapy for cancer patients [ 12 ] . DL, a sophisticated subset of ML, utilizes multilayer neural networks to analyze complex datasets and generate precise outcomes [ 13 , 14 ] . Key DL architectures include: Convolutional neural networks (CNNs): CNNs are specifically designed for image analysis, making them highly effective in medical imaging tasks. These networks use convolutional layers to detect patterns and features in images, such as tumors in radiological scans. For example, CNNs have achieved expert-level accuracy in detecting breast cancer from mammograms and identifying diabetic retinopathy in retinal images [ 15 ] . Recurrent neural networks (RNNs): RNNs are optimized for sequential data, such as time-series data or text. In healthcare, RNNs have been used to predict patient outcomes based on electronic health records (EHRs) and to analyze time-series data from wearable devices for early detection of health issues [ 16 ] . Convolutional neural networks (CNNs): CNNs are specifically designed for image analysis, making them highly effective in medical imaging tasks. These networks use convolutional layers to detect patterns and features in images, such as tumors in radiological scans. For example, CNNs have achieved expert-level accuracy in detecting breast cancer from mammograms and identifying diabetic retinopathy in retinal images [ 15 ] . Recurrent neural networks (RNNs): RNNs are optimized for sequential data, such as time-series data or text. In healthcare, RNNs have been used to predict patient outcomes based on electronic health records (EHRs) and to analyze time-series data from wearable devices for early detection of health issues [ 16 ] . Natural language processing (NLP), a branch of AI focused on understanding and processing human language, plays a critical role in healthcare [ 17 ] . NLP techniques include: Text classification: NLP models can classify clinical notes, patient records, and research articles into predefined categories. For example, NLP has been used to automatically categorize patient complaints in emergency departments, improving triage efficiency. Named entity recognition (NER): NER algorithms extract specific information, such as patient names, diagnoses, and medications, from unstructured text. This capability is essential for automating the extraction of relevant data from EHRs. Sentiment analysis: NLP can analyze patient feedback and social media posts to gauge public sentiment about healthcare services, providing valuable insights for improving patient care. Text classification: NLP models can classify clinical notes, patient records, and research articles into predefined categories. For example, NLP has been used to automatically categorize patient complaints in emergency departments, improving triage efficiency. Named entity recognition (NER): NER algorithms extract specific information, such as patient names, diagnoses, and medications, from unstructured text. This capability is essential for automating the extraction of relevant data from EHRs. Sentiment analysis: NLP can analyze patient feedback and social media posts to gauge public sentiment about healthcare services, providing valuable insights for improving patient care. Historically, rule-based expert systems, which rely on “if-then” rules, have seen commercial and healthcare applications, particularly in clinical decision support. These systems generate customized care plans based on patient assessments [ 18 ] . However, the scalability and adaptability of these systems are limited, especially when dealing with large rule sets or evolving knowledge domains. Consequently, there is a gradual shift toward data-driven and ML-based approaches in healthcare. In the realm of robotics, AI integration has enabled robots to perform predefined tasks such as hospital supply delivery and surgical assistance. These robots are evolving to be more collaborative and trainable. In surgical applications, while robots augment the capabilities of human surgeons, critical decision-making remains a human prerogative [ 19 ] . The integration of AI into robot operating systems is enhancing their functionality and adaptability in healthcare settings.

The future of AI in disease diagnostics is poised to revolutionize healthcare through advancements in technology, interdisciplinary collaboration, and innovative applications. Below, we outline key areas of progress and actionable steps to harness AI’s full potential. Robotic surgery, enhanced by AI and ML, is transforming minimally invasive procedures. The FDA-approved Da Vinci system has demonstrated efficacy in procedures such as robotic bronchoscopy for lung cancer diagnosis and endometriosis detection, offering improved precision and patient outcomes [ 57 , 58 ] . Future advancements should focus on: Developing autonomous robotic systems capable of real-time decision-making during surgery. Expanding the use of AI-powered robotics to underserved regions to improve access to advanced surgical care. Robotic surgery, enhanced by AI and ML, is transforming minimally invasive procedures. The FDA-approved Da Vinci system has demonstrated efficacy in procedures such as robotic bronchoscopy for lung cancer diagnosis and endometriosis detection, offering improved precision and patient outcomes [ 57 , 58 ] . Future advancements should focus on: Developing autonomous robotic systems capable of real-time decision-making during surgery. Expanding the use of AI-powered robotics to underserved regions to improve access to advanced surgical care. AI is revolutionizing cancer care by enabling personalized treatment strategies. ML models are being used to analyze tumor microenvironments, predict treatment responses, and identify novel biomarkers [ 59 ] . Future directions include: Integrating multi-omics data (genomics, proteomics, metabolomics) to refine risk assessment and treatment planning. Developing AI-driven platforms for real-time monitoring of cancer progression and therapy efficacy. Integrating multi-omics data (genomics, proteomics, metabolomics) to refine risk assessment and treatment planning. Developing AI-driven platforms for real-time monitoring of cancer progression and therapy efficacy. DP, powered by AI, is replacing traditional tissue biopsy with digital image analysis. AI algorithms have outperformed human pathologists in tasks such as detecting lymph node metastasis in breast cancer [ 60 – 62 ] . Future steps include: Standardizing DP workflows to ensure consistency and interoperability across healthcare systems. Leveraging AI to integrate molecular, histological, and immunohistological data for comprehensive diagnostic insights. Standardizing DP workflows to ensure consistency and interoperability across healthcare systems. Leveraging AI to integrate molecular, histological, and immunohistological data for comprehensive diagnostic insights. AI has shown remarkable accuracy in classifying skin lesions, with DL models achieving performance comparable to dermatologists [ 63 – 65 ] . Future advancements should focus on: Developing multimodal AI systems that combine imaging data with patient history and genetic information for enhanced accuracy. Deploying AI-powered diagnostic tools in primary care settings to improve early detection rates. Developing multimodal AI systems that combine imaging data with patient history and genetic information for enhanced accuracy. Deploying AI-powered diagnostic tools in primary care settings to improve early detection rates. The integration of AI with the IoT enables real-time monitoring and predictive analytics in healthcare. IoT sensors, combined with AI algorithms, can forecast patient health outcomes with high precision [ 66 – 68 ] . Future steps include: Developing secure IoT platforms to ensure data privacy and prevent cyberattacks. Expanding IoT applications to remote and resource-limited settings for decentralized healthcare delivery. Developing secure IoT platforms to ensure data privacy and prevent cyberattacks. Expanding IoT applications to remote and resource-limited settings for decentralized healthcare delivery. AI is enhancing genomics by identifying pathogenic variants and improving phenotype-to-genotype mapping [ 69 , 70 ] . Future directions include: Creating AI-driven platforms for personalized genomic medicine, enabling tailored treatments based on individual genetic profiles. Creating AI-driven platforms for personalized genomic medicine, enabling tailored treatments based on individual genetic profiles. Integrating genomic data with EHRs to provide holistic diagnostic insights. AI is streamlining healthcare operations by automating tasks such as scan interpretation and patient triage. For example, the UK’s National Health Service has implemented AI-powered chatbots for initial diagnosis, while the American College of Cardiology uses AI to reduce unnecessary imaging [ 71 , 72 ] . Future steps include: Developing AI-assisted decision support systems to reduce clinician workload and improve diagnostic accuracy. Implementing AI-driven tools to optimize resource allocation and reduce healthcare costs.

This narrative review was conducted to explore the role of AI in disease diagnostics and healthcare. A structured literature search was performed to identify relevant studies, focusing on advancements, applications, and challenges of AI in healthcare. The methodology was designed to provide a comprehensive overview rather than a systematic synthesis of evidence, as the aim was to contextualize the current state of AI in diagnostics and highlight key trends, gaps, and future directions. A systematic search was conducted across four major databases: PubMed, Scopus, Web of Science, and Google Scholar. The search terms included combinations of the following keywords: “artificial intelligence in healthcare” “ai in disease diagnostics” “machine learning in medical imaging” “deep learning in pathology” “ai in personalized medicine” “ethical challenges of ai in healthcare” “artificial intelligence in healthcare” “ai in disease diagnostics” “machine learning in medical imaging” “deep learning in pathology” “ai in personalized medicine” “ethical challenges of ai in healthcare” Boolean operators (AND, OR) were used to refine the search. The search was limited to peer-reviewed articles published in English up to January 2024. No restrictions were placed on study design to ensure a broad representation of AI applications in healthcare. The initial search yielded 1590 articles. After removing duplicates, 800 articles were screened based on their titles and abstracts. Studies were included if they: Discussed AI applications in healthcare, particularly in disease diagnostics. Provided comprehensive insights into AI technologies, their clinical applications, or associated challenges. Were published in English and peer-reviewed. Discussed AI applications in healthcare, particularly in disease diagnostics. Provided comprehensive insights into AI technologies, their clinical applications, or associated challenges. Were published in English and peer-reviewed. Studies were excluded if they: Focused on AI applications outside healthcare (e.g. finance, engineering). Lacked relevance to disease diagnostics or patient care. Were non-peer-reviewed or published in languages other than English. Focused on AI applications outside healthcare (e.g. finance, engineering). Lacked relevance to disease diagnostics or patient care. Were non-peer-reviewed or published in languages other than English. After full-text screening, 71 studies were selected for inclusion in this review. The selection process prioritized studies that offered novel insights, represented diverse applications of AI in healthcare, and addressed both technical and ethical challenges. This review adopts a narrative synthesis approach rather than a systematic review or meta-analysis, as the field of AI in healthcare is rapidly evolving with diverse applications and methodologies, making a broader exploration more suitable than strict quantitative synthesis. The included studies vary widely in design, focus, and methodology, which challenges systematic synthesis but are well-suited to a narrative approach that integrates diverse perspectives and findings. Additionally, the goal of this review is to provide a comprehensive overview of AI’s role in disease diagnostics, highlighting key advancements, applications, and challenges, rather than to answer a specific research question through quantitative analysis. While a Preferred Reporting Items for Systematic Reviews and Meta-Analyses flowchart is typically used for systematic reviews, it was not included here due to the narrative nature of this review. Instead, the study selection process is described in detail to ensure transparency. Given the narrative nature of this review, a formal quality assessment of individual studies (e.g. using tools like CASP or GRADE) was not conducted. However, studies were critically evaluated for their relevance, methodological rigor, and contribution to the field. Priority was given to high-impact studies published in reputable journals and those that provided robust evidence or novel insights.

The evolution of AI in healthcare can be contextualized within a rich historical framework. The inception of expert systems in healthcare aimed to replicate human cognitive processes. This concept was first articulated in 1950 with the Turing Test, which proposed computer simulations capable of intelligent behavior and critical thinking [ 7 ] . The practical application of AI in medicine was initially exemplified by the development of a glaucoma consultation program utilizing the CASNET model. This program offered personalized disease insights and expert advice to physicians, marking a significant milestone in AI’s medical applications [ 8 ] . A pivotal moment in the broader field of AI occurred in 1990 when IBM’s Deep Blue, a computer system programmed with advanced chess strategies, triumphed over Garry Kasparov, the reigning world chess champion. This achievement underscored AI’s potential in complex strategic thinking. Further advancements were seen in 2007 with the introduction of IBM’s Watson, an open-domain question-answering system. Watson’s application of DeepQA technology was a groundbreaking development in evidence-based clinical decision-making, demonstrating AI’s ability to process and analyze vast amounts of medical data [ 9 ] . With the advent of ML and DL, AI’s role in medicine has continued to evolve, opening avenues for personalized medicine and predictive models in disease diagnosis and therapeutic responses. However, the full potential of AI in healthcare remains to be thoroughly investigated and validated. Further research and additional clinical data are crucial to establish its efficacy and long-term value in medical practice.

Artificial intelligence (AI), an interdisciplinary domain of science and engineering, primarily focuses on creating and understanding intelligent computational processes and developing entities exhibiting such intelligence [ 1 ] . The proliferation of AI technologies has transformed various sectors, including healthcare, where its applications range from enhancing clinical decision-making to advancing drug development and personalized medicine. In the realm of healthcare, AI’s impact is profound and multifaceted, ranging from enhancing clinical decision-making to facilitating patient management of chronic diseases and advancing drug development [ 2 ] . Despite its transformative potential, the development and deployment of AI solutions in healthcare are fraught with challenges, notably high costs and complexity. AI’s capability to rapidly process and interpret complex clinical data – encompassing radiological, pathological, endoscopic, ultrasonographic, and biochemical assessments – is revolutionizing traditional medical diagnostics. Notably, AI algorithms have demonstrated superior performance in identifying malignancies in radiological images and optimizing cohort selection in clinical trials [ 2 ] . The utilization of AI in the medical field splits into two main areas: virtual and physical. Virtual AI covers machine learning (ML) and deep learning (DL), where DL is a branch of ML. These ML algorithms are divided into three types: supervised, unsupervised, and reinforcement learning, as illustrated in Figure 1 . On the other hand, the physical aspect of AI in medicine is represented by medical devices and robotics enhanced with AI technology. Figure 1. The difference between artificial intelligence (AI), machine learning (ML), and deep learning (DL). The difference between artificial intelligence (AI), machine learning (ML), and deep learning (DL). In the healthcare sector, AI has been effectively integrated across various key areas, demonstrating significant advancements and contributions. Clinical Decision Support Systems, which have been in use since the 1970s and 1980s, are designed to standardize medical practices and enhance adherence to established clinical guidelines [ 3 ] . Alongside this, Precision Medicine has emerged as a pivotal approach, utilizing individual genetic profiles to tailor treatment strategies and dosages, thereby personalizing patient care [ 4 ] . In the field of Image Analysis, AI has played a crucial role in reducing the variability in the interpretation of medical images. It has achieved a level of accuracy comparable to expert analysis in specialized areas such as retinal imaging and mammography [ 5 ] . Furthermore, the expanding domain of the Internet of Things (IoT), driven by the widespread adoption of smart devices and advances in sensor technology, has opened new and innovative avenues for applying AI within healthcare settings [ 6 ] . These integrations underscore the diverse and impactful role of AI in enhancing various facets of medical diagnostics and treatment. A critical gap in the current literature is the lack of comprehensive frameworks for addressing the ethical, legal, and societal implications of AI in healthcare. While numerous studies have explored AI’s technical capabilities, few have systematically examined its real-world applications and the challenges of implementing AI in diverse clinical settings, particularly in low-resource environments. Furthermore, there is limited research on how AI can be tailored to address global health disparities and ensure equitable access to its benefits. This narrative review aims to bridge these gaps by providing a comprehensive overview of AI’s role in disease diagnostics. It highlights key advancements, such as AI’s applications in cancer detection, drug discovery, and wearable health monitoring, while also addressing the ethical, technical, and regulatory challenges that must be overcome to realize AI’s full potential. By synthesizing current evidence and offering actionable recommendations, this review seeks to inform researchers, clinicians, and policymakers on the responsible integration of AI into healthcare systems.

The integration of AI into healthcare presents transformative opportunities but also raises significant ethical, legal, and societal challenges [ 40 ] . These challenges must be addressed to ensure that AI systems are safe, equitable, and aligned with the principles of patient-centered care (Fig. 2 ). Figure 2. Overview of various ethical and legal issues with using artificial intelligence in healthcare. Overview of various ethical and legal issues with using artificial intelligence in healthcare. AI’s application in healthcare introduces complex ethical questions that vary across regions and cultures. Core principles such as honesty, transparency, and respect for patient autonomy remain central, but their application to AI systems requires careful consideration. Key ethical issues include: Informed consent: Patients must be fully informed about the use of AI in their diagnosis and treatment. This includes understanding the limitations of AI, the potential for errors, and how their data will be used. For example, if an AI system is used to interpret medical images, patients should be aware that the results are probabilistic and may require human validation [ 41 ] . Transparency and explainability: Many AI systems, particularly DL models, operate as “black boxes,” making decisions without clear explanations. This lack of interpretability can undermine trust and hinder clinical adoption [ 42 ] . Explainable AI (XAI) techniques, such as saliency maps and decision trees, are being developed to address this issue, but further progress is needed to ensure that clinicians and patients can understand and trust AI-driven decisions. Algorithmic bias: AI systems can perpetuate or exacerbate existing biases if trained on unrepresentative datasets [ 43 ] . For instance, an AI model trained primarily on data from one demographic group may perform poorly for others, leading to disparities in care [ 44 ] . Addressing bias requires diverse and representative datasets, fairness-aware algorithms, and ongoing audits to identify and mitigate discriminatory outcomes [ 44 ] . Human empathy and job displacement: The rise of AI-driven tools, including robotic healthcare workers, raises concerns about the erosion of human empathy in patient care [ 45 ] . Additionally, the automation of tasks such as diagnostics and administrative work could lead to job displacement, necessitating workforce retraining and policy interventions to ensure a just transition. Informed consent: Patients must be fully informed about the use of AI in their diagnosis and treatment. This includes understanding the limitations of AI, the potential for errors, and how their data will be used. For example, if an AI system is used to interpret medical images, patients should be aware that the results are probabilistic and may require human validation [ 41 ] . Transparency and explainability: Many AI systems, particularly DL models, operate as “black boxes,” making decisions without clear explanations. This lack of interpretability can undermine trust and hinder clinical adoption [ 42 ] . Explainable AI (XAI) techniques, such as saliency maps and decision trees, are being developed to address this issue, but further progress is needed to ensure that clinicians and patients can understand and trust AI-driven decisions. Algorithmic bias: AI systems can perpetuate or exacerbate existing biases if trained on unrepresentative datasets [ 43 ] . For instance, an AI model trained primarily on data from one demographic group may perform poorly for others, leading to disparities in care [ 44 ] . Addressing bias requires diverse and representative datasets, fairness-aware algorithms, and ongoing audits to identify and mitigate discriminatory outcomes [ 44 ] . Human empathy and job displacement: The rise of AI-driven tools, including robotic healthcare workers, raises concerns about the erosion of human empathy in patient care [ 45 ] . Additionally, the automation of tasks such as diagnostics and administrative work could lead to job displacement, necessitating workforce retraining and policy interventions to ensure a just transition. The deployment of AI in healthcare also introduces complex legal challenges, particularly regarding accountability and liability. Key issues include: Accountability for errors: When an AI system makes an incorrect diagnosis or treatment recommendation, it is unclear who is responsible – the developer, the healthcare provider, or the institution using the technology [ 46 ] . For example, if an AI system fails to detect a tumor in a radiology scan, determining liability requires clear legal frameworks that define the roles and responsibilities of all stakeholders. Regulatory gaps: Existing regulations often lag behind the rapid pace of AI innovation. For instance, the U.S. FDA has established guidelines for AI-based medical devices, but these frameworks may not fully address the dynamic nature of AI systems that continuously learn and adapt [ 47 ] . Strengthening regulatory oversight and creating adaptive legal frameworks are essential to ensure patient safety. Intellectual property and data ownership: The use of patient data to train AI models raises questions about data ownership and intellectual property rights [ 48 ] . Clear policies are needed to balance the interests of patients, healthcare providers, and AI developers while ensuring that data usage complies with privacy laws such as the General Data Protection Regulation [ 49 ] . Accountability for errors: When an AI system makes an incorrect diagnosis or treatment recommendation, it is unclear who is responsible – the developer, the healthcare provider, or the institution using the technology [ 46 ] . For example, if an AI system fails to detect a tumor in a radiology scan, determining liability requires clear legal frameworks that define the roles and responsibilities of all stakeholders. Regulatory gaps: Existing regulations often lag behind the rapid pace of AI innovation. For instance, the U.S. FDA has established guidelines for AI-based medical devices, but these frameworks may not fully address the dynamic nature of AI systems that continuously learn and adapt [ 47 ] . Strengthening regulatory oversight and creating adaptive legal frameworks are essential to ensure patient safety. Intellectual property and data ownership: The use of patient data to train AI models raises questions about data ownership and intellectual property rights [ 48 ] . Clear policies are needed to balance the interests of patients, healthcare providers, and AI developers while ensuring that data usage complies with privacy laws such as the General Data Protection Regulation [ 49 ] . AI systems rely on vast amounts of sensitive patient data, making data security and privacy paramount. Key challenges include: Data breaches: AI systems are vulnerable to cyberattacks, which could compromise patient data or lead to harmful interventions [ 50 ] . For example, a hacked AI system could alter medical records or provide incorrect treatment recommendations, posing serious risks to patient safety [ 51 ] . Long-term data storage: Managing and securing large datasets over extended periods is a significant challenge. Ensuring data integrity and preventing unauthorized access requires robust encryption methods and secure storage solutions [ 52 ] . Synthetic data: To address privacy concerns, synthetic patient data – generated through AI to mimic real datasets – can be used for research and development [ 53 ] . This approach minimizes the risk of exposing sensitive patient information while enabling innovation [ 53 ] . Data breaches: AI systems are vulnerable to cyberattacks, which could compromise patient data or lead to harmful interventions [ 50 ] . For example, a hacked AI system could alter medical records or provide incorrect treatment recommendations, posing serious risks to patient safety [ 51 ] . Long-term data storage: Managing and securing large datasets over extended periods is a significant challenge. Ensuring data integrity and preventing unauthorized access requires robust encryption methods and secure storage solutions [ 52 ] . Synthetic data: To address privacy concerns, synthetic patient data – generated through AI to mimic real datasets – can be used for research and development [ 53 ] . This approach minimizes the risk of exposing sensitive patient information while enabling innovation [ 53 ] . Despite its advancements, AI cannot replace the human element in healthcare. Key limitations include: Trust and transparency: Patients and clinicians must trust AI systems to adopt them widely. Transparency in how AI models are developed, validated, and deployed is critical to building this trust [ 54 ] . Over-reliance on AI: Excessive dependence on AI could lead to deskilling among healthcare professionals and reduce their ability to make independent clinical judgments [ 55 ] . Balancing AI assistance with human expertise is essential to maintain high standards of care. Addressing health disparities: If not carefully designed, AI systems could worsen health disparities by favoring certain populations over others [ 56 ] . Ensuring equitable access to AI-driven healthcare requires proactive measures to address biases and promote inclusivity. Trust and transparency: Patients and clinicians must trust AI systems to adopt them widely. Transparency in how AI models are developed, validated, and deployed is critical to building this trust [ 54 ] . Over-reliance on AI: Excessive dependence on AI could lead to deskilling among healthcare professionals and reduce their ability to make independent clinical judgments [ 55 ] . Balancing AI assistance with human expertise is essential to maintain high standards of care. Addressing health disparities: If not carefully designed, AI systems could worsen health disparities by favoring certain populations over others [ 56 ] . Ensuring equitable access to AI-driven healthcare requires proactive measures to address biases and promote inclusivity.

The integration of AI into disease diagnostics has demonstrated transformative potential, offering unprecedented improvements in accuracy, efficiency, and personalization. From medical imaging to robotic surgery, AI is empowering healthcare professionals and enhancing patient outcomes. However, realizing AI’s full potential requires addressing key challenges and adopting actionable strategies. To enhance interpretability and transparency, developers must prioritize XAI models and establish guidelines for transparent reporting of AI algorithms. Addressing algorithmic bias and equity involves using diverse and representative datasets and implementing fairness-aware algorithms to ensure equitable healthcare delivery. Strengthening data security and privacy requires robust encryption methods, secure data storage solutions, and the use of synthetic data for research. Fostering interdisciplinary collaboration among technologists, clinicians, ethicists, and policymakers is essential to address ethical and legal challenges, while global standards for AI governance can ensure responsible implementation. The future of AI in disease diagnostics is bright, with advancements in robotic surgery, oncology, pathology, and IoT integration poised to transform healthcare.

The applications of AI in healthcare extend across multiple domains, each contributing to advancements in patient care, drug development, and outbreak prediction. Table 1 summarizes some of the key AI-driven healthcare applications, illustrating their purpose and providing references to relevant studies. Table 1 Overview of AI applications in healthcare: technologies, key benefits, and use cases. AI application Key purpose Technologies involved Key benefits References Disease diagnosis Early detection of diseases like cancer and neurological disorders. Deep learning, CNNs, NLP Improves diagnostic accuracy and enables early intervention. Memon et al [ 20 ] Drug discovery Accelerates identification of molecular patterns and drug efficacy. Machine learning, reinforcement learning Reduces time and cost in early drug development stages. Memon et al [ 20 ] Personalized treatment Tailors treatments based on genetic, lifestyle, and medical history data. Predictive analytics, genomic data integration Enhances treatment efficacy and reduces adverse effects. Chatterjee et al [ 21 ] Health records management Automates organization and retrieval of electronic health records. Natural language processing (NLP), blockchain Streamlines administration and ensures secure access to patient data. Luo et al [ 22 ] Clinical trial optimization Improves patient recruitment and study design using clinical data. Supervised learning, pattern recognition Enhances trial efficiency and reduces costs. Romanini et al [ 23 ] Crowd-sourced health data analytics Processes large-scale health data for research and public health monitoring. Big data analytics, cloud computing Enables epidemiological studies and improves public health responses. Rodrigues et al [ 24 ] Outbreak prediction Predicts and monitors disease outbreaks using multiple data sources. Machine learning, predictive models Increases epidemic preparedness and improves resource allocation. Chen et al [ 25 , 26 ] Medical imaging Enhance interpretation of medical images like X-rays and MRIs. CNNs, image processing Reduces diagnostic errors and supports early disease detection. Nasser et al [ 27 ] Overview of AI applications in healthcare: technologies, key benefits, and use cases. The table provides a snapshot of the diverse applications of AI in healthcare. AI’s ability to analyze complex datasets has proven especially effective in disease identification, drug development, and personalized medicine. Notably, its role in outbreak prediction and medical imaging diagnostics shows the potential to significantly impact public health and clinical outcomes. However, ongoing research is essential to address challenges like data privacy and algorithmic bias, ensuring that AI’s integration into healthcare continues to evolve responsibly.

The AI gets benefit from the EHR, health apps, and devices capable of being worn for detailed examination of the patterns and give actionable insight. ML algorithm is a good example which can predict the possible health risks on the basis of lifestyle data, giving power to timely intervention [ 28 ] . Moreover, Al-driven monitoring systems can detect anomalies in the vital signs so diminishing the errors of diagnosis and making the patient outcomes better [ 28 , 29 ] . AI’s first major medical use was in detecting atrial fibrillation. In 2014, AliveCor’s smartphone-based ECG monitor received Food and Drug Administration (FDA) approval for this purpose [ 30 ] . Also, I can expand on the challenges of false positives and their effects on patient care, as well as explore strategies to enhance elderly user engagement, such as kindly design and caregiver support. AI can also help in diagnosing endocrine diseases such as diabetes. Through DL, it can detect diabetic retinopathy from fundus images [ 31 ] . AI has benefits in detecting obstetric problems by enhancing ultrasound and MRI, helping to discover fetal anomalies and placental tumors. Based on AI algorithms, it improves fetal biometric measurements, which aid in monitoring fetal development [ 32 ] . In contrast to rheumatology, AI can help by relying on DL. It can also improve the interpretation of X-rays, MRIs, and ultrasounds to detect joint deformities and issues. AI algorithms can also highlight sacroiliac joint abnormalities using MRI [ 33 ] . In gastroenterology, AI, particularly CNNs, improves the detection of abnormalities in endoscopic and ultrasound images. In China, AI-assisted colonoscopies have shown a 20% higher polyp detection rate than traditional methods [ 34 ] . For example, Medtronic’s continuous glucose monitors help diabetes patients in managing blood sugar levels, assisting to reduce the stigma around hypoglycemia. However, such devices may sometimes cause anxiety for users [ 35 ] . We can enhance the nephrology example by incorporating additional details, such as specifying the AI models used (e.g. distinguishing between ML and DL), the datasets they are trained on, and the clinical validation processes. Moreover, providing references to relevant studies or real-world applications would better illustrate AI’s impact in this field. Let us know if this would enhance the section for you [ 36 ] . In gastroenterology, AI, particularly CNNs, improves the detection of abnormalities in endoscopic and ultrasound images [ 37 ] . Colonoscopies utilize AI in analyzing video feeds by ML detection of the potential polyps that unnoticed by the naked eye. Studies revealed that this applied science improves discovering rate by 20%, decreases human mistakes, and improve accuracy [ 37 ] . AI is also transforming pathology. Deep neural networks are used to automate biomarker assessments in breast tumors, improving the accuracy of classification, prognosis, and diagnosis [ 38 ] . This technology integrates demographic data, digital pathology (DP) images, and lab results to enhance treatment insights. For example, the Mayo Clinic uses AI for cervical cancer screening, achieving a 91% accuracy rate, which outperforms human experts at 69%. Despite challenges, pathologists across 59 countries agree that computational pathology will play a critical role in advancing the healthcare system [ 38 , 39 ] .