Impact of Artificial Intelligence on Healthcare: A Review of Current Applications and Future Possibilities

188 International Journal of Advance and Applied Research www.ijaar.co.in ISSN – 2347-7075 Impact Factor – 8.141 Peer Reviewed Bi-Monthly Vol. 6 No. 38 September - October - 2025 Impact of Artificial Intelligence on Healthcare: A Review of Current Applications and Future Possibilities Dr. Mukesh Tiwari1 & Dr. Khalid A. Shaikh2 1Department of Microbiology, Dr. D.Y.Patil ACS College Akurdi Pune. 2Department of Statistics, Dr. D.Y.Patil ACS College Akurdi Pune Corresponding Author –Dr. Mukesh Tiwari DOI - 10.5281/zenodo.17313161 Abstract: Artificial intelligence (AI) is rapidly transforming healthcare across diagnosis, treatment planning, patient monitoring, drug discovery, administration, and public health. This review synthesizes recent literature on AI applications in healthcare, highlights concrete examples where AI has influenced clinical practice (medical imaging, diabetic retinopathy screening, protein-structure prediction for drug discovery), examines benefits (accuracy, efficiency, scalability), and details persistent challenges (data bias, interpretability, regulatory and ethical concerns). We also survey regulatory progress and guidance, and outline likely near-term and long-term future possibilities, including personalized medicine driven by multi-modal models, AI-assisted clinical trials, and integration of large language models (LLMs) into clinical workflows. Finally, we propose priorities for research, governance, and safe deployment to maximize benefit while minimizing harms. Key policy and scientific developments indicate AI’s promise but underline the need for robust regulation, transparency, and emphasis on equity. Keywords: Artificial Intelligence, Machine Learning, Healthcare, Medical Imaging, Drug Discovery, Ethics, Regulation, Large Language Models, Alphafold. Introduction: Artificial intelligence (AI) — the application of machine learning (ML), deep learning (DL), and related computational techniques — has rapidly shifted from research prototypes to deployed systems in clinical settings. Applications range from diagnostic imaging and clinical decision support to drug discovery and administrative automation. The COVID-19 pandemic accelerated interest and deployment of AI tools for surveillance, diagnosis, and resource planning. However, the field also faces challenges: algorithmic bias, questions about clinical generalizability, data governance, explainability, and the need for regulatory frameworks to ensure patient safety and equitable benefit. Healthcare is undergoing a profound transformation driven by the rapid integration of digital technologies, with Artificial Intelligence (AI) emerging as one of the most influential forces of change. AI refers to the capability of computer systems to simulate human cognitive functions such as learning, reasoning, and decision-making. In healthcare, these technologies are not only supporting clinicians in diagnostic accuracy and therapeutic planning but are also reshaping how patients engage with health services, how hospitals manage resources, and how research is conducted. Unlike conventional digital tools, AI systems can process vast amounts of IJAAR Vol. 6 No. 38 ISSN – 2347-7075 Dr. Mukesh Tiwari &Dr. Khalid A. Shaikh 189 complex data, recognize subtle patterns, and generate actionable insights with a level of efficiency and consistency that surpasses traditional approaches. The increasing adoption of AI in healthcare has been motivated by multiple factors. The exponential growth of health data from electronic medical records, imaging modalities, genomics, wearable devices, and remote monitoring systems has created a pressing need for advanced analytical solutions. At the same time, healthcare systems worldwide are facing significant challenges, including rising costs, a shortage of skilled professionals, aging populations, and the growing burden of chronic diseases. AI technologies offer innovative solutions to address these challenges by enhancing clinical decision support, optimizing hospital operations, improving early disease detection, and enabling personalized medicine. Current applications of AI in healthcare are diverse and span almost every aspect of the care continuum. Machine learning algorithms are increasingly applied to diagnostic imaging, pathology, and dermatology, where they assist in detecting abnormalities with remarkable accuracy. Natural language processing facilitates the interpretation of unstructured medical notes, streamlining record-keeping and clinical documentation. AI-enabled predictive models are proving valuable in identifying patients at risk of complications, readmission, or disease progression, thus allowing for proactive interventions. Robotics, powered by AI, are enhancing surgical precision and rehabilitation outcomes, while conversational agents and chatbots are extending support in patient engagement, telemedicine, and mental health care. Despite these advancements, the integration of AI into healthcare also brings forth several challenges and ethical considerations. Issues such as data privacy, algorithmic bias, lack of transparency in decision-making (often termed the ―black box‖ problem), and regulatory uncertainty remain significant obstacles to widespread adoption. Furthermore, the human aspect of care— empathy, trust, and communication—cannot be replaced by technology and needs to be preserved even as AI systems become more prevalent. Ensuring equitable access to AIdriven healthcare across different regions and populations is also a critical concern, particularly in lowand middle-income countries where healthcare disparities already exist. Looking ahead, the future possibilities of AI in healthcare are both exciting and complex. With advancements in deep learning, precision medicine, and integration with other emerging technologies such as blockchain and the Internet of Medical Things (IoMT), AI has IJAAR Vol. 6 No. 38 ISSN – 2347-7075 Dr. Mukesh Tiwari &Dr. Khalid A. Shaikh 190 the potential to create a more efficient, predictive, and patient-centered healthcare ecosystem. However, realizing this vision requires careful balancing of innovation with ethical responsibility, multidisciplinary collaboration between clinicians, technologists, and policymakers, and robust frameworks for governance and regulation. This review seeks to provide a comprehensive examination of the current applications of AI in healthcare and to explore the future possibilities that could redefine the practice of medicine. By analyzing the benefits, limitations, and ethical implications, the paper aims to contribute to a deeper understanding of how AI can be harnessed to improve health outcomes while maintaining the core values of patient care. Methods — literature review approach: This review used a targeted narrative synthesis of peer-reviewed journals, major institutional reports, and regulatory announcements published through 2025. Databases and sources included PubMed/PMC, Nature and Science family journals, World Health Organization (WHO) guidance, and U.S. Food and Drug Administration (FDA) documents. Searches combined terms such as ―AI in healthcare review,‖ ―machine learning medical imaging,‖ ―AI drug discovery,‖ ―WHO AI health guidance,‖ and ―FDA AI medical devices.‖ Recent high-impact examples (e.g., AlphaFold, DeepMind ophthalmology systems) and regulatory actions were prioritized to illustrate major trends and policy responses. This is a qualitative review intended to synthesize contemporary evidence and identify gaps and directions for future work . Current Applications: 1. Medical Imaging and Diagnostics: AI has arguably achieved its earliest clinical impact in medical imaging (radiology, pathology, ophthalmology). Convolutional neural networks (CNNs) and related architectures have been used to detect fractures, lung nodules, breast lesions on mammograms, and retinal pathology on fundus photos and OCT scans. Real-world evaluations show AI systems can reach sensitivity and specificity comparable to specialists in narrow tasks, and can reduce reading time. Notable examples include DeepMind/Moorfields work on OCT interpretation (comparable accuracy to ophthalmologists) and multiple FDAauthorized AI imaging tools for triage and detection. However, real-world performance depends critically on data distribution match and robust external validation. IJAAR Vol. 6 No. 38 ISSN – 2347-7075 Dr. Mukesh Tiwari &Dr. Khalid A. Shaikh 191 2. Electronic health records (EHR) and clinical decision support: Natural language processing (NLP) and predictive models applied to EHR data are used to predict patient deterioration (sepsis, ICU transfer), assist in medication dosing, and automate documentation. LLMs and specialized medical language models can summarize notes, draft discharge summaries, and extract structured data from free text. While these tools can reduce clinician administrative burden, concerns about hallucinations, data privacy, and reliability in clinical reasoning persist. 3 Drug discovery and molecular biology: AI tools accelerate several stages of drug discovery: target identification, molecule generation and optimization, and proteinstructure prediction. AlphaFold (DeepMind) has transformed structural biology by predicting protein 3-D structures at an accuracy that rivals experiments for many proteins; its impact on target characterization and rational drug design is substantial. AIdriven generative chemistry and virtual screening platforms are being used to propose candidate molecules, though translating AI outputs into clinically approved drugs remains challenging and resource-intensive. NatureScienceDirect 4. Genomics and precision medicine: AI models synthesize genomic, transcriptomic, and clinical data for risk stratification and to suggest therapy options (e.g., tumor genomics driving targeted therapies). Multi-omic integration using machine learning supports biomarker discovery and personalized therapeutic approaches, though reproducible validation and access to representative datasets are ongoing needs. 5. Robotics, surgery, and procedural assistance: Surgical robots (e.g., da Vinci) combined with AI for motion analysis, augmented reality overlays, and automated suturing research are evolving. These systems are enhancing minimally invasive procedures, training, and intraoperative decision support, but fully autonomous surgical robots remain experimental. 6. Telemedicine, remote monitoring, and wearables: AI algorithms process continuous data from wearables (ECG, activity trackers) for arrhythmia detection, fall risk prediction, and chronic disease monitoring. Telehealth platforms increasingly incorporate AI triage and symptom checkers to support remote consultations, expanding access but raising questions about accuracy and liability. 7. Administrative and operational tasks: AI streamlines billing, coding, appointment scheduling, and resource optimization. These applications can reduce costs and free clinician time but also raise workforce and fairness concerns. Representative case studies / Examples:  Diabetic retinopathy screening: Multiple deep-learning systems have demonstrated specialist-level detection of referable diabetic retinopathy on retinal images and have been trialed in screening programs to increase coverage where specialist access is limited. PMC  Ophthalmic OCT interpretation (DeepMind & Moorfields): Systems detecting a wide range of retinal pathologies from OCT images matched clinician performance in studies, illustrating how imaging AI can triage and prioritize patients. STAT  AlphaFold for structural biology: AlphaFold’s accurate protein structure predictions have accelerated molecular IJAAR Vol. 6 No. 38 ISSN – 2347-7075 Dr. Mukesh Tiwari &Dr. Khalid A. Shaikh 192 understanding and informed structurebased drug design workflows. Nature Benefits and opportunities:  Improved diagnostic accuracy and speed: AI can detect subtle patterns and process large image volumes faster than humans in narrow tasks.  Scalability: Once validated, AI tools can be deployed at scale — particularly valuable in low-resource settings (e.g., AI screening programs).  Personalization: ML models that integrate multi-modal data (imaging + genomics + EHR) support individualized risk stratification and treatment planning.  Drug discovery acceleration: AI reduces early-stage candidate space exploration time and can suggest novel chemistries and targets using protein structure predictions.  Operational efficiencies: Automation of documentation and administrative workflows can free clinician time for patient care. Challenges, risks, and ethical considerations: Data quality, representativeness, and bias: Models trained on biased or unrepresentative datasets risk poor generalizability and can perpetuate health disparities. Sourcing diverse, high-quality labeled datasets and transparent reporting of dataset composition are essential. PMC Explainability and trust: Many high-performing models (deep neural networks, LLMs) lack transparent reasoning pathways. Clinicians and patients may distrust systems that cannot explain decisions; explainable AI (XAI) research aims to bridge this gap but is not a universal solution. Safety, validation, and regulatory oversight: Clinical safety requires rigorous prospective evaluation and continuous monitoring post-deployment because model performance can degrade over time as clinical practice and data distributions change. Regulators worldwide are issuing guidance for AI/ML medical devices; the FDA has issued and refined guidance for lifecycle management and marketing of AI-enabled devices. Robust premarket evidence and postmarket surveillance are increasingly emphasized. U.S. Food and Drug Administration Exponent Privacy and data governance: Use of patient data in training large models raises consent, de-identification, and data-sharing concerns. National and international policy frameworks must balance innovation with privacy protection. Liability and professional roles: Responsibility for AI-assisted decisions — clinician, developer, or institution — needs clear legal frameworks. Changes to clinical roles and training are necessary if AI takes over routine tasks. Hallucination and misinformation (for LLMs): Large language models can produce plausible but incorrect outputs (―hallucinations‖), which is hazardous in clinical contexts unless properly constrained and validated. Regulation, governance, and ethics — recent developments: International bodies and regulators have begun issuing guidance to govern health AI. The WHO has published ethical and governance recommendations for AI in health emphasizing equity, transparency, and human oversight. National regulators, including the FDA, have developed guidance on AI/ML in medical devices, addressing premarket evidence, cybersecurity, and IJAAR Vol. 6 No. 38 ISSN – 2347-7075 Dr. Mukesh Tiwari &Dr. Khalid A. Shaikh 193 predetermined change control plans for adaptive systems. These developments reflect recognition that AI requires lifecycle oversight: from data provenance and model evaluation to post-market monitoring and handling of model updates. World Health Organization U.S. Food and Drug Administration Future Possibilities (near-term and longterm): Near-term (1–5 years):  Broad integration of AI triage and decision support into EHR workflows — automated alerts for deterioration, medication interactions, and documentation assistance.  LLM-powered documentation and information retrieval — reducing clinician administrative burden while requiring guardrails to prevent hallucinations.  AI-assisted imaging triage deployed in screening programs to expand access in underserved regions.  Improved model monitoring and ―model-as-medical-device‖ practices — continuous validation pipelines, data drift detection, and safer model updates under regulatory frameworks. 8.2 Mid/long-term (5–15+ years)  Personalized, multi-modal AI clinicians: integrated systems that combine genomics, continuous physiologic monitoring, imaging, and social determinants to deliver tailored care plans.  AI-accelerated drug pipelines: computational platforms that significantly shorten lead discovery and optimize molecules—potentially reducing costs and timelines for certain drug classes, but still requiring clinical trials. AlphaFold and successor models will underpin many design workflows. PMC  Assistive autonomy in certain procedures: semi-autonomous robots for repetitive surgical tasks under human supervision.  Population-level public health surveillance: AI systems for early outbreak detection, resource allocation forecasting, and epidemic control. Research & policy priorities: To enable safe and equitable benefits from AI in healthcare, priorities include: 1. High-quality, diverse datasets with clear provenance and standardized reporting. 2. Prospective, multi-center clinical trials and real-world evidence for AI tools, not only retrospective validation. 3. Interoperability standards and transparent evaluation metrics for comparing algorithms across settings. 4. Robust regulatory frameworks that address adaptive systems, post-market surveillance, and transparency requirements. 5. Ethics and fairness auditing with stakeholder participation, especially from communities at risk of being disadvantaged by biased models. 6. Clinician training and role redesign to work with AI, including how to interpret outputs and manage failure modes. World Health OrganizationExponent Limitations of current evidence: Many AI studies remain single-center, retrospective, or use convenience datasets that overstate real-world performance. Heterogeneous reporting standards make cross-study comparisons difficult. The rapid pace of AI innovation also means that reviews can become quickly dated; continuous IJAAR Vol. 6 No. 38 ISSN – 2347-7075 Dr. Mukesh Tiwari &Dr. Khalid A. Shaikh 194 synthesis and living systematic reviews are needed to keep guidance current. PMC Conclusion: AI presents transformative potential for healthcare: improved diagnostics, novel drug discovery workflows, operational efficiencies, and expanded access to care. However, realizing these benefits equitably and safely demands rigorous validation, transparent governance, and global coordination on ethics and regulation. Advances such as AlphaFold demonstrate AI’s power to solve previously intractable scientific problems, while regulatory moves from bodies like the WHO and FDA show policymakers are adapting to the technology’s realities. The future landscape will likely feature hybrid systems where human clinicians and AI cooperate — combining human judgment, values, and oversight with machine scalability and pattern recognition. For policymakers, clinicians, and researchers, the challenge is to accelerate innovation while embedding safeguards that protect patients and ensure equitable distribution of benefits. Nature World Health Organization Acknowledgment: The authors gratefully acknowledge financial support from the Dr. D. Y. Patil ACS College, Akurdi, Pune, Maharashtra. References: 1. Alhejaily AMG. Artificial intelligence in healthcare (Review). PMC 2024. PMC 2. Al Kuwaiti A, et al. A Review of the Role of Artificial Intelligence in Healthcare. PMC 2023. PMC 3. World Health Organization. Harnessing artificial intelligence for health / WHO guidance on AI in health. (WHO digital health & innovation pages). World Health Organization+1 4. U.S. Food and Drug Administration. Artificial Intelligence and Machine Learning in Software as a Medical Device (FDA pages; draft/final guidance). 2024– 2025 updates. U.S. Food and Drug AdministrationExponent 5. Jumper J, Evans R, Pritzel A, et al. Highly accurate protein structure prediction with AlphaFold. Nature 2021. Nature 6. Desai D. Review of AlphaFold 3: Transformative Advances in Drug Design and Therapeutics. PMC 2024/2025. PMC 7. Kong M, et al. Artificial Intelligence Applications in Diabetic Retinopathy. PMC 2024. PMC 8. Khalifa M, et al. AI in diagnostic imaging: Revolutionising accuracy and ... 2024. ScienceDirect/Nature review. ScienceDirect 9. Goodman KW, et al. Global health and big data: The WHO's artificial ... (Ethics & governance reviews). PMC 2023. PMC 10. Chustecki M. Benefits and Risks of AI in Health Care: Narrative Review. Interact J Med Res. 2024;13(1):e53616. IJMR 11. Khan MM, et al. Towards Secure and Trusted AI in Healthcare: A Systematic Review. (2010–2023). Comput Methods Programs Biomed. 2024; e-pub. ScienceDirect 12. Botha NN, et al. Perceived Threats Posed by the Use of AI Tools in Healthcare on Patients’ Rights and Safety: A Scoping Review (2010–2023). Arch Public Health. 2024; e-pub. BioMed Central 13. Morone G, et al. Artificial Intelligence in Clinical Medicine: Overview of Systematic Reviews. Front Digit Health. 2025; e-pub. Frontiers 14. Awasthi R. Artificial Intelligence in Healthcare: 2024 Year in Review. medRxiv. 2025; e-pub. MedRxiv 15. De Micco F, et al. Artificial Intelligence in Healthcare: Transforming Patient Clinical Risk Management. Front Med. 2025; e-pub.