Artificial intelligence in modern clinical practice

Luis Inglada Galiana ¹, Francisco Javier Mena Martin ², Pablo Anselmo Sanz Espinosa ³

¹ Department Internal Medicine in Valladolid. Grupo Recoletas
² Chief of Internal Medicine Department, Universitary Hospital Rio Hortega, Valladolid 47012
³ Resident of Internal Medicine, Universitary Hospital Rio Hortega in Valladolid

[email protected]

OPEN ACCESS

PUBLISHED 31 August 2025

CITATION Inglada Galiana, L., Mena Martin, F.J., Sanz Espinosa, P.A., 2025. Artificial intelligence in modern clinical practice. Medical Research Archives, [online] 13(8). https://doi.org/10.18103/mra.v13i8.6887

COPYRIGHT © 2025 European Society of Medicine. This is an open-access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited.

DOI https://doi.org/10.18103/mra.v13i8.6887

ISSN 2375-1924

ABSTRACT

Overview: Artificial Intelligence (AI) has transformed from theoretical concept to practical reality in healthcare, revolutionizing disease diagnosis, treatment, and management. This technology uses machine learning and deep learning algorithms to analyze complex medical datasets, significantly improving diagnostic accuracy, treatment efficiency, and personalized patient care.

Clinical Applications: AI has revolutionized medical imaging across specialties. In radiology, systems detect lung nodules and pneumonia with high accuracy, while supporting mammography for early cancer detection. Digital pathology benefits from AI’s ability to identify cancers and quantify biomarkers invisible to human eyes. Ophthalmology and dermatology applications include detecting diabetic retinopathy and classifying skin lesions with specialist-level accuracy. Beyond imaging, AI enables early disease detection by integrating electronic health records and biomarkers to identify predictive patterns before symptoms appear. Applications span oncology risk prediction, cardiovascular ECG analysis, and chronic disease management through wearable device monitoring.

Treatment and Operations: AI transforms treatment through personalized medicine, combining genomic and clinical data to predict therapy responses. In surgery, AI enhances robot-assisted procedures with real-time feedback and precision guidance. Drug discovery acceleration includes genomic database analysis and virtual compound screening, with AI-developed drugs entering clinical trials. Healthcare operations benefit from AI through intelligent scheduling, patient flow management, and resource allocation. Natural Language Processing extracts valuable information from clinical documentation, while predictive analytics optimize hospital workflows and supply chain management.

Challenges and Future Directions: Despite promising applications, AI faces significant implementation challenges. Algorithmic bias risks perpetuating healthcare disparities, while “black box” models limit transparency and clinical trust. Data privacy, regulatory frameworks, integration costs, and clinician resistance present additional barriers. The future lies in collaborative models where AI enhances rather than replaces clinical expertise. Success requires coordinated efforts to develop explainable, robust systems while addressing ethical concerns and ensuring equitable implementation that maintains core healthcare values.

Three Principal Keywords: Machine Learning – Personalized Medicine – The ultimate goal of tailoring treatments to individual patient characteristics Clinical D.

Artificial intelligence in modern clinical practice

Artificial Intelligence often abbreviated as AI refers to systems or machines capable of performing tasks that traditionally require human intelligence, such as learning from experience, solving problems, recognizing patterns, and making decisions. What once seemed like a futuristic idea is now a powerful force shaping many areas of society, especially healthcare.

In recent years, AI has moved beyond theoretical discussions and entered real-world clinical settings, offering the potential to transform how diseases are diagnosed, treated, and managed. From algorithms that detect subtle changes in medical images to systems predicting drug responses, AI is becoming an essential tool for improving diagnostic accuracy, treatment efficiency, and personalized patient care.

This transformation has been fueled by the explosion of health data from electronic records and imaging to genomics and wearable devices. Machine learning, particularly deep learning using artificial neural networks, has proven remarkably effective at analyzing these complex datasets. These tools are helping address major challenges such as chronic disease management, aging populations, healthcare disparities, and resource limitations.

Given its rapid evolution and wide-ranging impact, it’s crucial to take a close look at AI’s role in modern medicine. While there’s excitement about its potential, there are also important concerns around implementation, limitations, and ethical implications. Healthcare professionals, researchers, policymakers, and the public need a clear understanding of where AI stands today, what works, what doesn’t, and how to move forward responsibly. This review aims to offer a comprehensive overview of AI in clinical practice.

We’ll explore AI’s contributions across key areas: diagnosis, treatment planning, patient monitoring, hospital operations, and personalized medicine through genomic and proteomic analysis. Beyond clinical applications, we’ll examine how AI affects healthcare administration such as optimizing workflows and analyzing electronic health records through natural language processing (NLP).

A significant part of this article will focus on the ethical, legal, and social issues surrounding AI in medicine. Topics include algorithmic bias, data privacy, accountability in decision-making, transparency, and regulatory challenges related to fair and safe use.

Our goal is to provide a balanced and thorough analysis that highlights both technological advancements and the critical considerations needed for responsible integration of AI into clinical settings with the ultimate aim of improving patient outcomes and well-being.

This article draws from an extensive review of published scientific literature including original research, reviews, meta-analyses, and relevant clinical guidelines from reputable databases. The most recent and impactful studies were prioritized to ensure an up-to-date, well-supported perspective on this evolving field.

Materials & Methods

A comprehensive search strategy was adopted for this methodological review to include data from diverse, recent, and most cited sources of studies.

Search strategy

Data were collected via literature search through various online sources including Google Scholar, PubMed, NIH (National Library of Medicine), Web of Science, European database, Springer, and Embase databases. Since the study focused on applications of artificial intelligence in medicine and healthcare, the major research items were of things and medical terms, among other similar search terms. After a thorough analysis of titles and abstracts of publications related to applications of AI in the medicine and healthcare industry, the data was selected for this review study. Only studies published in the English language are included in this study. Moreover, it is tried that only the data from 2019 onwards be included in the study.

Inclusion and exclusion criteria

The study has not been restricted to just research articles; rather, multiple types of sources have been used, including book chapters, review articles, case reports, clinical trials, and case studies that have been published since 2019. Approximately, more than 75 studies have been conducted to extract the findings related to applications of artificial intelligence in healthcare and medicine.

AI in clinical diagnosis

AI has proven to be an incredibly powerful tool in clinical diagnosis, thanks to its ability to analyze massive, complex datasets and detect subtle patterns that might otherwise go unnoticed. This section explores how AI is being applied in medical imaging, early disease detection, and personalized medicine through genomic and proteomic analysis.

Applications in Medical Imaging

Medical imaging covering radiology, pathology, ophthalmology, and dermatology has seen some of the most impressive breakthroughs in AI. Interpreting images is often complex, and even experienced professionals can face challenges due to variability or subtle anomalies. Deep learning models, particularly convolutional neural networks (CNNs), have demonstrated performance levels that match and sometimes exceed those of human experts.

In radiology, AI plays a key role in detecting abnormalities such as lung nodules or signs of pneumonia in chest X-rays, with high sensitivity and specificity. It also supports mammography by identifying suspicious lesions, improving early cancer detection while reducing the burden on radiologists. AI helps segment organs, measure tumor volumes, and detect strokes in CT and MRI scans, acting like a second pair of eyes to improve diagnostic accuracy and streamline workflow.

Digital pathology has been revolutionized by the digitization of tissue samples. AI assists in identifying and grading cancers such as prostate or breast cancer by detecting microscopic features invisible to the human eye. It also quantifies biomarkers like PD-L1, which are crucial for targeted therapies, enhancing both diagnostic consistency and efficiency.

In ophthalmology, AI detects visual disorders such as diabetic retinopathy, macular degeneration, and glaucoma with accuracy comparable to specialists. This capability is especially valuable for large-scale screening programs aimed at preventing irreversible vision loss.

In dermatology, AI classifies skin lesions like malignant melanoma using clinical or dermatoscopic images, performing at levels similar to experienced dermatologists. Mobile apps based on AI are emerging for preliminary self-assessment, particularly useful in areas where specialist access is limited.

Early Disease Detection

Beyond imaging, AI plays a vital role in catching diseases at their earliest stages often before symptoms appear. By integrating electronic health records, biomarkers, and sensor data, AI identifies predictive patterns that can lead to earlier intervention.

In oncology, AI builds risk prediction models based on genetic, environmental, and lifestyle factors, helping identify individuals at high risk. It’s also used in liquid biopsies, analyzing tumor biomarkers in blood for non-invasive cancer detection.

For cardiovascular disease, AI analyzes ECGs to detect arrhythmias like atrial fibrillation in asymptomatic patients. Predictive models assess the likelihood of events like heart attacks or heart failure, guiding preventive strategies.

In neurology, AI seeks to detect early signs of Alzheimer’s and Parkinson’s by analyzing brain imaging, cerebrospinal fluid, speech patterns, motor function, and digital behavior. Although still under study, these approaches hold promise for earlier, more accurate diagnosis.

Genomics, Proteomics, and Personalized Medicine

The explosion of molecular data has created a demand for AI to interpret genetic and protein information and advance personalized medicine. In genomics, AI identifies disease-related genetic variants, predicts mutation effects, and analyzes gene expression profiles. In cancer, it helps select targeted therapies like immunotherapy based on tumor genetics and aids in diagnosing rare diseases using genome/exome data.

Proteomics and metabolomics also benefit from AI. It analyzes proteins or metabolites in body fluids to find biomarker patterns, enabling new diagnostic tests for conditions like cancer, infections, or autoimmune diseases.

Combining multi-omic and clinical data through AI allows the creation of personalized health models sometimes called “digital twins” that simulate individual physiology and predict treatment response. Although widespread adoption remains challenging, this marks a major step toward precision medicine.

Conclusion Diagnostic Section

AI is reshaping clinical diagnosis from improving image interpretation and enabling early disease detection to integrating molecular data into personalized medicine. However, full integration into clinical practice requires rigorous validation and careful attention to ethical, regulatory, and implementation challenges. Collaboration among developers, clinicians, and researchers is essential to ensure these tools deliver real benefits to patients.

AI in treatment planning and execution

Once a diagnosis has been made, Artificial Intelligence (AI) opens up new possibilities for planning and delivering treatment. By analyzing vast amounts of clinical data, medical literature, and trial results, AI helps clinicians make more informed, precise, and personalized therapeutic decisions. It is also reshaping surgery and interventional procedures, while speeding up drug discovery and development.

Personalized Treatment Recommendation Systems

Personalized medicine tailoring treatments to individual patients is one of the most promising frontiers in healthcare, and AI plays a central role. By combining genomic, clinical, and lifestyle data, AI models can predict how different patients will respond to various therapies, allowing for more accurate treatment selection.

In oncology, AI analyzes tumor profiles and patient histories to recommend the most effective immunotherapies or targeted treatments. This helps oncologists choose from an expanding array of options and biomarkers to maximize effectiveness while minimizing side effects. Platforms like IBM Watson for Oncology have attempted this, though real-world adoption has faced challenges due to limitations in performance and integration into clinical workflows.

For chronic conditions such as diabetes or cardiovascular disease, AI supports more tailored care. For example, it can forecast glycemic responses to different foods or adjust insulin dosing based on continuous glucose monitoring and activity levels. In cardiology, AI aids decision-making regarding revascularization strategies or predicts bleeding risk under various antithrombotic treatments.

These systems are designed to support not replace clinical judgment. They help doctors consider multiple variables simultaneously, improving decision-making without removing the human element. Transparency and explainability remain key to gaining clinician trust and ensuring safe, effective use.

Robotic Surgery and AI-Guided Interventions

AI is also transforming surgical procedures through advanced robotics and intelligent navigation tools. These technologies enhance precision, reduce complications, and improve outcomes across a range of specialties.

In robot-assisted surgery, AI enhances visualization by overlaying critical information such as vessel locations or tumor margins onto the surgical field. It can analyze intraoperative images and robotic data in real time, offering feedback, preventing errors, or even automating repetitive tasks. Micro-surgical tools powered by AI can reduce hand tremors or assist in delicate suturing. Research is ongoing into partially autonomous robotic systems that could perform certain procedures with minimal human intervention, although ethical and safety concerns remain.

In interventional radiology, AI assists in planning procedures such as tumor ablation or embolization by optimizing paths that avoid critical structures. It also fuses preoperative imaging with real-time fluoroscopy or ultrasound to guide interventions more accurately, reducing procedure time and radiation exposure.

In radiotherapy, AI automates the segmentation of organs at risk and delineation of tumor volumes on CT or MRI scans. It generates highly targeted treatment plans that deliver maximum tumor dose while sparing healthy tissue. Adaptive radiotherapy, which adjusts treatment plans in real time based on anatomical changes, is increasingly supported by AI.

AI-Accelerated Drug Discovery and Development

Developing new drugs is a lengthy, expensive, and high-risk process. AI is emerging as a powerful tool to streamline every phase from identifying therapeutic targets to designing clinical trials and repurposing existing medications.

AI analyzes genomic, proteomic, and bibliographic databases to uncover disease-related proteins and pathways, predicting which might be “druggable.” This helps prioritize research efforts and speeds up early-stage discovery.

It also screens millions of virtual compounds using deep learning models to estimate physicochemical properties, biological activity, and toxicity narrowing down the list of potential candidates for testing. During preclinical and clinical phases, AI improves trial design by selecting patients based on predictive biomarkers and analyzing real-time data for safety signals or efficacy trends. It also facilitates drug repurposing by identifying new uses for existing medications through molecular profiling.

Pharmaceutical and biotech companies are heavily investing in AI, with several AI-developed drugs already entering clinical trials. While experimental validation remains crucial, AI has the potential to significantly shorten development timelines and reduce costs.

Conclusion Treatment Section

AI is redefining how treatment is planned and delivered from personalizing therapy selection and enhancing surgical precision to accelerating drug development. However, realizing its full potential requires continued research, rigorous validation, and responsible implementation guided by ethical standards and collaboration among developers, regulators, and healthcare professionals.

AI in patient monitoring and management

Artificial Intelligence (AI) has opened new possibilities for continuous patient monitoring by analyzing real-time data streams and identifying subtle patterns that may signal health changes. This capability is transforming how both acute and chronic conditions are managed whether in hospitals or at home. From interpreting data from wearable devices to predicting disease progression and optimizing intensive care, AI offers tools that make healthcare more proactive, personalized, and efficient.

Remote Patient Monitoring and Wearable Device Data Analysis

The rise of wearable sensors and mobile health technologies has led to an explosion of physiological data such as heart rate, sleep quality, oxygen saturation, blood pressure, and glucose levels collected in everyday settings. AI plays a crucial role in converting this raw information into meaningful insights, enabling more responsive and effective remote patient monitoring (RPM).

For example, in managing heart failure, AI can analyze weight fluctuations, blood pressure trends, and reported symptoms to predict impending decompensation, allowing earlier intervention and potentially preventing hospitalization. In patients with chronic obstructive pulmonary disease (COPD), AI detects shifts in breathing patterns and reduced physical activity that may indicate an approaching exacerbation.

One of the key strengths of AI is its ability to personalize alerts and recommendations. Instead of relying on generic thresholds, AI models learn each patient’s baseline and flag only significant deviations. This reduces false alarms and increases the accuracy of interventions. By integrating data from multiple sources including environmental and behavioral factors AI provides a more comprehensive view of a patient’s condition.

Beyond chronic diseases, AI-assisted RPM is proving valuable in post-operative recovery, mental health management (e.g., detecting sleep disturbances or activity changes linked to depressive episodes), and promoting healthy behaviors through tailored feedback.

Predictive Analytics for Chronic Disease Outcomes and Management

AI-driven predictive analytics holds great promise in managing long-term conditions, which place a heavy burden on global healthcare systems. By analyzing extensive historical clinical datasets including electronic health records, lab results, genomic profiles, and socioeconomic factors AI models can forecast disease onset, progression, and the risk of complications or adverse events.

In diabetes, machine learning algorithms help anticipate risks for complications such as nephropathy, retinopathy, or neuropathy, enabling early and targeted preventive measures. These models also support individualized treatment plans by forecasting how patients might respond to different therapies. For those already diagnosed, AI optimizes glycemic control by predicting glucose trends and recommending adjustments in diet, physical activity, or medication.

For cardiovascular diseases, AI identifies patients at high risk of sudden cardiac events like those who might benefit from defibrillators and predicts the progression of atherosclerosis. This allows for better risk stratification and more efficient allocation of healthcare resources.

In chronic respiratory conditions like asthma or COPD, AI uses lung function measurements, medication adherence records, symptom reports, and environmental factors (e.g., pollen or pollution levels) to forecast flare-ups. These predictions enable clinicians to take proactive steps and reduce unnecessary emergency visits.

A particularly critical application is early detection of sepsis in intensive care units (ICUs). Sepsis is life-threatening and requires immediate action. AI continuously monitors clinical and laboratory data, identifying early signs of deterioration hours before traditional methods potentially improving outcomes by enabling faster treatment.

Applications in Mental Health and Wellbeing

AI is increasingly being used in mental health, both for monitoring and delivering interventions. Given the shortage of mental health professionals in many regions and the stigma associated with seeking help, AI-based tools offer promising ways to improve access and quality of care.

For monitoring, AI analyzes diverse data sources to detect early signs of mental illness or relapse. Natural Language Processing (NLP) techniques allow AI to interpret social media posts, text messages, or digital diary entries for language patterns linked to depression, anxiety, or suicide risk. Smartphone sensors can track behavioral indicators like activity levels, phone usage, and social interactions to provide additional insights into emotional well-being.

While these innovations raise important ethical questions particularly around privacy and consent they open up new opportunities for early detection and preventive care.

For therapeutic use, AI-powered chatbots are emerging as virtual mental health assistants. These tools deliver cognitive behavioral therapy exercises, mindfulness practices, and coping strategies, offering immediate and confidential support. While they are not substitutes for professional care, they can be especially helpful for people with mild to moderate symptoms or as complements to traditional treatments.

AI also enables personalized digital interventions by tailoring content and timing to individual needs. By analyzing clinical, genetic, and neuroimaging data, AI helps predict treatment responses, paving the way for more precise psychiatric care.

AI in healthcare operations and administration

Beyond diagnosis and treatment, AI is playing an increasingly important role in optimizing the operations and administration of healthcare systems. The complexity of managing hospitals, clinics, and health networks along with the need to improve efficiency, reduce costs, and ensure quality provides fertile ground for AI-based solutions. This section explores how AI is helping to enhance workflows, analyze clinical and administrative data, and manage healthcare resources more intelligently.

Workflow Optimization and Hospital Efficiency

Healthcare institutions are complex environments with numerous interconnected processes that can greatly benefit from AI-driven optimization. Workflow inefficiencies lead to delays, increased costs, and staff burnout. AI tools help analyze processes, identify bottlenecks, and suggest improvements.

One key application is optimizing appointments and outpatient capacity. AI algorithms predict no-show rates and optimize scheduling to reduce wait times and maximize use of clinical resources. By analyzing historical patterns, these systems implement smart overbooking strategies or send personalized reminders to reduce absences.

Within hospitals, AI improves patient flow management from admission to discharge. Predictive models anticipate emergency department volumes, length of stay, and bed allocation needs. This helps reduce congestion and improve discharge planning. It also supports operating room optimization, considering procedure duration, staff availability, and urgency maximizing efficiency and reducing idle time.

AI also enhances supply chain management. Algorithms predict demand for pharmaceuticals and medical devices, optimizing inventory to avoid shortages or surpluses and improving distribution logistics critical during crises like pandemics.

Furthermore, AI can reduce administrative errors and improve documentation. Automated systems verify diagnostic and procedural coding (e.g., ICD-10 or CPT) for billing, reducing mistakes and improving reimbursement.

Natural Language Processing (NLP) for Clinical Documentation and Electronic Health Records (EHRs)

Electronic Health Records (EHRs) contain valuable clinical information, much of it in unstructured text such as progress notes or pathology reports. NLP, the AI branch focused on human language interaction, is essential for extracting value from this data.

One NLP application is enhancing clinical documentation. Voice recognition systems enable efficient dictation of notes, and NLP algorithms structure this information or extract key clinical concepts relieving administrative burden and improving EHR quality.

NLP also facilitates analysis of large clinical datasets for research and quality improvement. Algorithms extract data on diagnoses, symptoms, medications, or outcomes enabling cohort creation, adverse event monitoring, and clinical guideline adherence evaluation. It helps identify at-risk patients and monitor healthcare-associated infections.

Another growing application is improving communication between patients and providers. AI-powered chatbots answer frequently asked questions, assist with appointment scheduling, and provide medical information freeing up clinical staff time. NLP can also analyze doctor-patient communication to identify improvement opportunities or assess patient understanding.

NLP’s ability to transform unstructured text into usable data is critical for other AI applications such as decision support systems, predictive analytics, and translational research.

Resource Management and Demand Forecasting

Efficient resource management personnel, equipment, facilities, and budgets is a constant challenge. AI provides tools to plan and allocate resources more intelligently by forecasting service demand.

AI models can predict future demand from hours to years ahead. These range from emergency department traffic to long-term care needs of aging populations, using historical, demographic, epidemiological (e.g., flu outbreaks), and external data (e.g., weather or large events). Such predictions enable proactive staffing and capacity adjustments.

Regarding staff, AI can optimize scheduling for physicians and nurses by accounting for workload variations, staff preferences, and skill requirements improving satisfaction, reducing burnout, and ensuring adequate coverage.

It also applies to expensive medical equipment like MRI or CT scanners. AI predicts when maintenance is needed (predictive maintenance), reducing downtime and extending service life. Usage is also optimized for better operational efficiency.

From a public health perspective, AI supports strategic resource allocation in preventive and health promotion programs. By identifying at-risk populations or high-need areas, interventions can be targeted where they will have the greatest impact.

Conclusion Operations and Administration Section

AI is transforming healthcare system management by offering powerful tools to improve efficiency, quality, and sustainability. By optimizing workflows, extracting value from clinical and administrative data, and enabling more predictive and proactive resource management, its proper implementation has the potential to revolutionize healthcare operations. Achieving this requires careful integration with existing systems, staff training, and continuous evaluation of impact and value.

Ethical, legal, and social implications (ELSI) of AI in clinical practice

The accelerated integration of Artificial Intelligence (AI) into clinical practice marks a major step forward in medicine, offering the potential to enhance diagnostics, improve treatment strategies, and optimize healthcare administration. However, this transformation also brings complex ethical, legal, and social implications that must be carefully addressed. Public trust and successful adoption of these technologies depend heavily on how well these challenges are managed. This section explores key issues such as algorithmic bias, data privacy, accountability for AI-driven decisions, and the need for appropriate regulatory frameworks.

Algorithmic Bias and Equity in Healthcare

One of the most pressing ethical concerns surrounding AI in medicine is the risk of reinforcing or even amplifying existing health disparities. Since AI models learn from historical data, any biases present such as differences in care based on race, gender, or socioeconomic status can become embedded in their outputs.

For instance, an AI trained on non-representative datasets may perform poorly for underrepresented groups, leading to inaccurate diagnoses or missed opportunities for early intervention. Cases have been documented where facial recognition systems showed higher error rates for darker-skinned individuals or women raising concerns if similar tools are used in dermatology or remote monitoring. Likewise, models designed to estimate disease risk or healthcare needs can reflect past inequalities in access to care or coding practices if not properly audited.

To address these risks, a comprehensive approach is necessary: ensure training data is diverse and representative; implement tools to detect and reduce algorithmic bias; regularly audit model performance across different population groups; maintain transparency regarding data sources and evaluation metrics; and encourage diversity within development teams to incorporate broader perspectives and reduce blind spots.

Privacy and Security of Health Data

AI relies on vast amounts of highly sensitive health information, raising serious concerns about data privacy and security. Breaches or misuse of this data could lead to discrimination, identity theft, or stigmatization. The challenge is heightened when data is shared across institutions for training purposes, increasing exposure to potential vulnerabilities. While anonymization techniques are commonly used, they are not foolproof.

To mitigate these risks: strong cybersecurity measures such as encryption, access controls, and regular audits should be standard practice; emerging approaches like federated learning where models are trained across multiple decentralized sites without sharing raw data and differential privacy which introduces statistical noise to protect identities are promising solutions; robust legal protections, including frameworks like the General Data Protection Regulation (GDPR) in Europe and HIPAA in the U.S., are essential for safeguarding patient rights; and clear informed consent processes and transparency around AI use must be central to implementation efforts.

The World Health Organization (WHO) has also issued guiding principles aimed at ensuring AI serves the public interest globally. These emphasize fairness, transparency, and global cooperation in developing and deploying AI responsibly.

Accountability, Transparency, and Explainability of AI Models

Determining responsibility when AI makes an error remains a challenging legal issue. If an AI-driven recommendation leads to a harmful outcome, who is held accountable the developer, the hospital using the system, or the clinician who followed its suggestion? Existing legal frameworks, which were built around traditional medical errors, struggle to keep pace with AI’s complexity. This issue is further complicated by the “black box” nature of many deep learning models, which makes it difficult to understand how decisions are made, limiting clinicians’ ability to validate or question them.

To address this: greater emphasis is being placed on Explainable AI (XAI), which aims to develop models that can justify their decisions in ways that clinicians can understand highlighting key variables or providing natural language explanations; even if full interpretability isn’t always possible, a minimum level of transparency is crucial for building trust, enabling clinical validation, and supporting accountability; and clinicians must retain final decision-making authority. AI should support not replace professional judgment and the contextual understanding that comes with medical expertise.

Regulatory Challenges and Governance Frameworks

Regulating AI in healthcare presents unique difficulties. It requires balancing innovation with patient safety and ethical considerations. Traditional regulatory models designed for pharmaceuticals and medical devices are not always suitable for adaptive AI systems that continuously evolve. Agencies like the FDA in the U.S. and EMA in Europe are adapting their approaches to evaluate AI-based software, including requirements for real-world testing and ongoing monitoring after deployment.

Beyond product-level regulation, broader governance structures are needed at both institutional and societal levels. These should: incorporate clear ethical guidelines; support research into the ethical, legal, and social impacts of AI; involve diverse stakeholders including patients, clinicians, developers, and the general public; and invest in educating both healthcare providers and AI developers on responsible use and safe implementation.

Given the global reach of AI innovation, international collaboration will be vital to align standards, share best practices, and establish common ethical and regulatory principles applicable across different healthcare systems.

Conclusion ELSI Section

While AI holds immense potential to transform medicine, its responsible integration depends on proactively addressing regulatory, ethical, and social challenges. Reducing bias, protecting data privacy, clarifying accountability, and building robust governance structures are essential steps toward ensuring that AI supports fair, safe, and beneficial clinical applications.

Discussion: advances, challenges, and future directions of AI in clinical practice

Artificial Intelligence has emerged as one of the most promising and disruptive technologies in modern medicine, with the potential to fundamentally reshape how healthcare is delivered in the 21st century. Throughout this review, we’ve explored its impact across multiple domains from improving diagnostic accuracy through advanced imaging and genomic analysis to supporting personalized treatment planning, enhancing surgical precision, accelerating drug discovery, and optimizing patient monitoring.

The common thread across these applications is AI’s ability to process and interpret vast, complex datasets revealing insights and patterns that augment clinical decision-making and lead to more accurate, timely, and tailored care. Advances in machine learning, particularly deep learning techniques, have enabled algorithms to achieve performance levels comparable to or even exceeding those of human experts in specific tasks.

In diagnostics, AI-powered tools have significantly improved the interpretation of radiological images, pathology slides, and retinal scans enabling earlier detection of conditions like cancer and diabetic retinopathy. In treatment, clinical decision support systems are helping oncologists tailor therapies based on tumor genetics, while robotic-assisted surgery and adaptive radiotherapy are pushing the boundaries of procedural precision.

Drug development is another area where AI is making waves. By analyzing massive datasets and screening virtual compound libraries, AI is speeding up target identification and candidate selection, potentially reducing both time and cost in bringing new medications to market. Meanwhile, predictive analytics powered by AI are transforming chronic disease management by anticipating complications before they occur, allowing for early intervention.

Beyond direct patient care, AI is also streamlining healthcare operations. From workflow optimization and hospital resource allocation to natural language processing of electronic health records, AI-driven solutions are helping institutions improve efficiency and reduce administrative burdens.

Together, these innovations promise a future of more accurate diagnoses, fewer clinical errors, reduced clinician workload, and broader access to specialized medical knowledge.

Current Limitations of Clinical AI

Despite these advancements, integrating AI into real-world clinical settings comes with significant challenges. One of the most pressing issues is the “black box” nature of many AI models, especially those built using deep learning. The lack of transparency makes it difficult for clinicians to understand how decisions are reached, which limits trust, complicates validation, and raises concerns about accountability.

Another major limitation is generalizability. AI models trained on data from one population or institution often struggle when applied elsewhere due to differences in demographics, clinical protocols, or equipment. Without rigorous, multicenter validation in real-world environments, widespread deployment remains risky.

Data quality and availability also pose obstacles. Medical data can be fragmented, incomplete, or biased leading to flawed predictions if not carefully curated. Preparing high-quality, representative datasets requires substantial effort and resources, yet it’s essential for building reliable AI systems.

Most current AI applications are narrow in scope, designed for specific, well-defined tasks. While effective in controlled settings, these models often lack the contextual reasoning and adaptability needed to handle the complexity and variability of real-world clinical practice.

Barriers to Widespread Clinical Adoption

Even with strong technical capabilities, AI adoption in healthcare faces structural and cultural barriers. Integration into existing clinical workflows is far from seamless. Many AI tools fail to interoperate smoothly with electronic health record systems or disrupt established routines, making them impractical for daily use.

Cost is another limiting factor. Developing, implementing, and maintaining AI solutions can be expensive particularly for smaller or under-resourced institutions. These costs go beyond software and hardware; they include training staff, adapting workflows, and ensuring ongoing support.

Resistance from clinicians is also a reality. Some professionals express concern over losing autonomy, fearing that reliance on AI may depersonalize care or create dependency on unfamiliar digital tools. Overcoming this skepticism will require robust evidence of AI’s safety and effectiveness, along with comprehensive education and training programs.

Regulatory and reimbursement frameworks remain underdeveloped in many regions. The absence of clear guidelines on approval, validation, and coverage for AI-based services hinders both innovation and implementation.

Adding to these hurdles are the ethical, legal, and social implications of AI especially around algorithmic bias, data privacy, and responsibility for automated decisions. Addressing these concerns is crucial for ensuring safe, equitable, and trustworthy integration into clinical practice.

Conclusions and Future Directions

For AI to fulfill its potential in medicine, a coordinated research agenda must address current limitations. This includes developing models that are more explainable, robust, and broadly applicable across diverse populations and settings. Improving data quality and representativeness is equally important, as is creating strategies that allow AI tools to integrate effectively into clinical workflows.

Clear regulatory frameworks and governance structures must also be established to promote responsible innovation while safeguarding core healthcare values. Training healthcare professionals to work alongside AI and involving patients and other stakeholders in the process will be essential for building trust and ensuring that AI contributes equitably and ethically to better health outcomes.

As AI continues to evolve, future developments are likely to focus on multimodal systems capable of integrating diverse data sources and delivering more context-aware insights. Rather than replacing human expertise, the most successful applications will likely emerge from collaboration where AI supports and enhances clinical judgment rather than substituting for it.

With thoughtful development, rigorous evaluation, and ethical implementation, AI holds the potential to revolutionize medicine not just by improving diagnostics and treatments, but by reshaping how care is delivered and experienced at every level.

Edge computing

Edge computing in medicine involves processing data closer to where it is generated, like medical devices or wearables, rather than relying on centralized cloud servers. This approach is transforming healthcare by enabling faster, more efficient, and secure data handling. A breakdown of its key impacts based on current insights: is a distributed computing paradigm that processing data closer to where it is generated such as medical devices, wearables, or IoT sensors rather than relying on centralized cloud servers. In medicine, this means faster data analysis, reduced latency, and improved privacy for applications like real-time diagnostics, patient monitoring, and AI-driven analytics.

Examples:

• A smart insulin pump with edge computing can monitor glucose levels, adjust dosages in real-time, and alert doctors to anomalies without cloud dependency.

Real-Time Diagnostics and Monitoring:

Edge computing allows devices like wearable heart monitors or portable imaging systems to analyze data instantly. For example, ECG patches can detect arrhythmias on the spot, alerting patients or doctors without delay. This reduces latency compared to sending data to the cloud, which can take seconds or more critical in emergencies like stroke or cardiac events.

Data Privacy and Security:

By processing sensitive patient data locally (e.g., on an edge server), edge computing minimizes the need to transmit data over the internet, reducing exposure to breaches. This aligns with regulations like HIPAA, ensuring compliance while handling electronic health records (EHRs) or genomic data.

Remote and Resource-Limited Settings:

In rural clinics or ambulances, where internet connectivity is spotty, edge devices enable diagnostics without cloud reliance. For instance, portable ultrasound machines with edge AI can analyze scans locally, supporting faster decision-making in areas lacking specialists.

AI Integration:

Edge computing powers AI models on devices for tasks like image analysis (e.g., detecting tumors in X-rays) or predictive analytics (e.g., forecasting seizures from EEG data). This reduces bandwidth costs and enables offline functionality, crucial for low-resource environments.

Challenges:

Despite benefits, edge computing faces hurdles like high initial costs for infrastructure, interoperability issues between devices, and the need for robust cybersecurity to protect edge nodes. Scalability can also be tricky, as managing numerous edge devices across a hospital network requires sophisticated orchestration.

Examples in Action:

Posts on X highlight edge role in wearable health monitors and real-time analytics for patient care. Web sources note its use in smart hospitals, where IoT devices manage everything from patient vitals to equipment maintenance, cutting response times and costs.

General Conclusion

Artificial Intelligence has become one of the most promising and transformative technologies in modern clinical practice, with the potential to fundamentally reshape how healthcare is delivered in the 21st century. Throughout this review, we’ve explored a wide range of AI applications from improving diagnostic accuracy through advanced analysis of medical images and genomic data, to supporting personalized treatment planning, enhancing surgical procedures, and accelerating drug discovery. We’ve also examined how AI is transforming continuous patient monitoring, chronic disease management, and the operational efficiency of healthcare institutions.

At the heart of these innovations lies AI’s ability to process and interpret vast, complex datasets revealing patterns and insights that enhance clinical decision-making and lead to more accurate, proactive, and personalized care. Advances in machine learning, particularly deep learning, have enabled algorithms to achieve expert-level performance in specific tasks. The promise is a healthcare model that is not only more effective against disease but also more equitable, efficient, and tailored to individual patient needs. AI also offers the opportunity to reduce clinicians’ workload, allowing them to focus more on direct patient interaction and the human aspects of medicine.

However, the integration of AI into clinical settings is not without challenges. Its success depends on addressing limitations such as transparency, explainability, robustness, and generalizability. The quality and representativeness of the data used to train these models are crucial, and there is a real risk of reinforcing or amplifying existing disparities if biases are not properly addressed. Ethical, legal, and social implications particularly around data privacy, accountability for algorithmic decisions, and regulatory frameworks are complex and require ongoing, multidisciplinary dialogue.

Widespread adoption of AI also faces practical barriers, including seamless workflow integration, high implementation costs, and the need for proper training of healthcare professionals. Overcoming these obstacles will require sustained investment not only in AI technology itself but also in real-world evaluation and understanding its impact on patients, providers, and healthcare systems.

Looking ahead, AI is expected to continue evolving rapidly, with the emergence of more sophisticated, multimodal models capable of contextual reasoning. The most likely and desirable path forward is one of collaboration where AI enhances rather than replaces clinical expertise and judgment. For this future to benefit everyone equitably, AI development and implementation must be guided by strong ethical principles, a commitment to fairness, and an unwavering dedication to patient safety and well-being.

In the end, artificial intelligence is not a cure-all, but a powerful tool whose impact depends entirely on how we design, implement, and govern it. The promise of AI-transformed healthcare is immense but its realization will require wisdom, foresight, and collaborative effort to ensure that this technological revolution serves to improve the human condition and promote health for all.

Modern clinical practice stands at the threshold of a new era and artificial intelligence is, without a doubt, at its core.

Conflict of Interest Statement:

None.

Funding Statement:

None.

Acknowledgements:

To all institutions, organizations, and individuals who contribute to the advancement and dissemination of artificial intelligence its foundations and applications through online platforms and scientific journals, with particular emphasis on the field of medicine.

References:

1. Topol EJ. High-performance medicine: the convergence of human and artificial intelligence. Nat Med 2019;25(1):44-56.
2. Jiang F, Jiang Y, Zhi H, Dong Y, Li H, Ma S et al. Artificial intelligence in healthcare: past, present and future. Stroke Vasc Neurol 2017;2(24):230-243.
3. Rajpurkar P, Chen E, Banerjee O, Topol EJ. AI in health and medicine. Nat Med 2022;28(1):31-38.
4. Inglada Galiana L, Corral Gudino L, Miramontes González P. Ethics and artificial intelligence. Rev Clin Esp (Barc) 2024;224(3):178-186. doi: 10.1016/j.rceng.2024.02.003
5. Bohr A, Memarzadeh K. The rise of artificial intelligence in healthcare applications. In: Adam Bohr, Kaveh Memarzadeh. Artificial Intelligence in Healthcare. Academic Press; 2020. p. 25-60. https://doi.org/10.1016/B978-0-12-818438-7.00002-2.
6. Obermeyer Z, Emanuel EJ. Predicting the Future Big Data, Machine Learning, And Clinical Medicine. N Engl J Med 2016;375(13):1216-1219. doi: 10.1056/NEJMp1606181.
7. Esteva A, Robicquet A, Ramsundar B, Kuleshov V, DePristo M, Chou C, et al. A guide to deep learning in healthcare. Nat Med 2019;25(1):24-29. doi: 10.1038/s41591-018-0316-z. Epub 2019 Jan 7.
8. Panch T, Pearson-Stuttard J, Greaves F, Atun R. Artificial intelligence: opportunities and risks for public health. Lancet Digit Health 2019;1(1):e13-e14. doi: 10.1016/S2589-7500(19)30002-0
9. Wiens J, Saria S, Sendak M, Ghassemi M, Liu VX, Doshi-Velez F et al. Do not harm: a roadmap for responsible machine learning for health care. Nat Med 2019;25(9):1337-1340. doi: 10.1038/s41591-019-0548-6
10. Kelly CJ, Karthikesalingam A, Suleyman M, Corrado G, King D. Key challenges for delivering clinical impact with artificial intelligence. BMC Med 2019;17(1):195. doi: 10.1186/s12916-019-1426-2.
11. McKinney SM, Sieniek M, Godbole V, Godwin J, Antropova N, Ashrafian H et al. International evaluation of an AI system for breast cancer screening. Nature 2020;577(7788):89-94. doi: 10.1038/s41586-019-1799-6.
12. Ardila D, Kiraly AP, Bharadwaj S, Choi B, Reicher JJ, Peng L et al. End-to-end lung cancer screening with three-dimensional deep learning on low-dose chest computed tomography. Nat Med 2019;25(6):954-961. doi: 10.1038/s41591-019-0447-x
13. Bera K, Schalper KA, Rimm DL, Velcheti V, Madabhushi A. Artificial intelligence in digital pathology new tools for diagnosis and precision oncology. Nat Rev Clin Oncol 2019;16(11):703-715. doi: 10.1038/s41571-019-0252-y
14. Ting DSW, Cheung CYL, Lim G, Tan GSW, Quang ND, Gan A et al. Development and validation of a Deep Learning System for Diabetic Retinopathy and Related Eye Diseases Using Retinal Images From Multiethnic Populations With Diabetes. JAMA 2017;318(22):2211-2223. doi: 10.1001/jama.2017.18152
15. Alpar S, AC. Evaluating smartwatch-based detection of supraventricular tachycardia and atrial fibrillation in the emergency department. Am J Emerg Med 2025;95:101-106.
16. Lin E, Lin CH, Lane HY. Deep Learning with Neuroimaging and Genomics in Alzheimer’s Disease. Int J Mol Sci. 2021;22(15):7911.
17. Cohen JD, Li L, Wang Y, Thoburn C, Afsari B, Danilova L et al. Detection and localization of surgically resectable cancers with a multi-analyte blood test. Science 2018;359:926-930. doi:10.1126/science.aar3247
18. Christopoulos G, Graff-Radford J, Lopez CL, Yao X, Attia ZI, Rabinstein AA et al. Artificial Intelligence-Electrocardiography to Predict Incident Atrial Fibrillation: A Population-Based Study. Circ Arrhythm Electrophysiol 2020;13(12):e009355. doi: 10.1161/CIRCEP.120.009355
19. Sahu A, Nema P, Rajak D, Purohit A, Rawal R, Soni V, Kashaw SK. Exploring AI tools and multi-omics for precision medicine in lung cancer therapy. Cytokine Growth Factor Rev 2025:S1359-6101(25)00071-1. doi: 10.1016/j.cytogfr.2025.06.001.