ARTIFICIAL INTELLIGENCE IN HEALTHCARE: DIAGNOSTIC SUPPORT AND ADMINISTRATIVE AUTOMATION
Authors:
- Julia ŻakJŻJulia Żak
- Monika StęPińskaMSMonika StęPińska
- Marta OmiecińskaMOMarta Omiecińska
- Alicja MaciejewskaAMAlicja Maciejewska
- Maja KaczorMKMaja Kaczor
- Weronika TrynkiewiczWTWeronika Trynkiewicz
- Zuzanna RybkaZRZuzanna Rybka
- Emilia LenkiewiczELEmilia Lenkiewicz
- Karolina DąbrowskaKDKarolina Dąbrowska
- Jakub WiniarczykJWJakub Winiarczyk
Open Access
- 12 May 2026
- journal article
- Published by RS Global Sp. z O.O. in International Journal of Innovative Technologies in Social Science
- Vol. 1 (2(50))
Abstract
Background: Artificial intelligence (AI) is rapidly moving from theoretical research into daily hospital operations. While AI promises to improve diagnostic speed and reduce administrative burnout, its real-world success depends on how well it integrates into complex clinical workflows. This review evaluates the practical impact of both discriminative and generative AI on modern healthcare delivery. Methods: We conducted a targeted literature search using PubMed and Semantic Scholar (via Consensus AI) for research published between January 2018 and March 2026. We structured this work as a critical narrative review to focus specifically on the practical challenges and successes of AI in real-world hospital settings. After screening for clinical relevance and the presence of deployment data, we selected 21 peer-reviewed articles for final inclusion. Results: The evidence shows that discriminative AI significantly improves speed in acute care, particularly in sepsis forecasting and stroke triage. Newer models are also showing success in evaluating surgical risks from ECGs and tracking traumatic brain injuries. However, these tools often face challenges like "alert fatigue" and data bias. On the administrative side, generative AI—such as ambient clinical scribes—is successfully reducing documentation time and improving billing accuracy. We found that the success of these tools depends on "Human-in-the-Loop" models and mathematical explainability frameworks, like SHAP values, which help doctors understand and trust AI decisions. Conclusion: AI is becoming a necessary requirement for modern medicine, not just an optional upgrade. By offloading high-volume data tasks to machines, AI can help return the physician’s focus to direct patient care. Future success depends on using protocols that ensure algorithmic transparency and prevent automation bias, ensuring the doctor remains the final authority in clinical decisions.Keywords
This publication has 21 references indexed in Scilit:
- Generative AI in preanesthetic consultations: Effects on efficiency, documentation workload, quality, and physician–patient interaction: A simulation trialJournal of Clinical Anesthesia, 2026
- Evaluation of an Artificial Intelligence Defined Lung Nodule Malignancy Score in Incidental Pulmonary Nodules: The CREATE StudyMayo Clinic Proceedings: Digital Health, 2026
- Clinician Experiences With Ambient Scribe Technology to Assist With Documentation Burden and EfficiencyJAMA Network Open, 2025
- Generative Artificial Intelligence Use in Healthcare: Opportunities for Clinical Excellence and Administrative EfficiencyJournal of Medical Systems, 2025
- Large language models encode clinical knowledgeNature, 2023
- Algorithmic fairness in artificial intelligence for medicine and healthcareNature Biomedical Engineering, 2023
- Prospective, multi-site study of patient outcomes after implementation of the TREWS machine learning-based early warning system for sepsisNature Medicine, 2022
- Predicting cancer outcomes with radiomics and artificial intelligence in radiologyNature Reviews Clinical Oncology, 2021
- Deep learning for chest X-ray analysis: A surveyMedical Image Analysis, 2021
- How artificial intelligence could transform emergency department operationsThe American Journal of Emergency Medicine, 2018