ºÚÁÏÕýÄÜÁ¿

Katsuri Kirubaharan

Author's Statement

This essay was written in a section of 76-101 Interpretation and Argument with Dr. Rochel Gasson to explore the integration of artificial intelligence in healthcare, a topic deeply meaningful to me as I am majoring in business while pursuing the pre-med track. The assignment challenged me to engage in academic inquiry, construct a persuasive argument, and consider multiple perspectives on a complex, real-world issue, all of which were objectives of this course. My goal was to analyze how AI can enhance patient care, increase workflow efficiency, and personalize treatment, while also acknowledging potential negatives such as ethical concerns and accessibility gaps.

Through this work, I practiced adapting different academic genres and styles to connect with readers, refining my argumentation, and situating my research within an ongoing scholarly conversation. Composing this essay strengthened my ability to critically evaluate evidence, structure reasoned arguments, and communicate complex ideas clearly. On a personal level, this assignment reinforced the importance of healthcare in my life and how technology will continue to both advance and challenge patient care.

- Kasturi Kirubaharan

In what ways is artificial intelligence reshaping clinical care, administration, and organizational practices in Pittsburgh’s healthcare sector?

Abstract

This essay examines how artificial intelligence is reshaping clinical care, administrative operations, and organizational practices within Pittsburgh’s healthcare system. As AI becomes more integrated into medicine, it influences multiple areas of medicine, including diagnostics, data processing, and healthcare system functioning. Although there is existing research on AI’s benefits in diagnostics, workflow efficiency, and operational effectiveness, there are also challenges of AI relating to bias, inequity, and regulation. This essay focuses primarily on clinical performance rather than the systematic or healthcare administrative side. This essay contributes to analyzing AI as a technical and organizational resource, showing how its adoption affects patient care, along with staffing, resource allocation, and decision-making effectiveness. The methodology that will be used includes surveys, interviews, and case studies within the UPMC system. By integrating quantitative and qualitative data, the essay identifies patterns in how AI shapes professional responsibilities of healthcare professionals, patient engagement, and systemic workflow. This essay investigates how healthcare professionals perceive AI’s usefulness, risks, ethical implications, and training requirements. Understanding AI’s multidimensional impact offers insight into how healthcare systems can create human-centered, equitable, and responsibly regulated AI integration strategies, specifically for improving patient outcomes and addressing healthcare disparities across regions such as medical deserts.

Introduction

The integration of Artificial Intelligence (AI) into healthcare has led to the introduction of many arguments about the ethical effects of AI and how it should be regulated. AI, or artificial intelligence, which is defined as the science and engineering of making intelligent machines that mimic human decision-making and problem-solving, is being increasingly applied to clinical and operational healthcare processes. Before exploring the ethics of AI use in healthcare, however, it is important to understand some crucial language within the field. Some important definitions include machine learning, which is training algorithms to mimic human decision-making; AI fairness, which is mitigating bias in AI to promote equity; and AI bias, which is systemic preferences or disadvantages in AI outcomes often reflecting human biases. Some other terms related to health which are important to understand include health equity, which is ensuring equal opportunity to achieve full health potential, and medical deserts, which are areas with limited access to healthcare. Understanding these definitions helps to accurately characterize flaws within the system, like how racial bias has been seen in clinical risk prediction algorithms as many clinicians underestimate the severity of illnesses and diseases in Black patients. Similarly, gender bias can be seen in diagnostic tools, since they are usually made from male data, which often results in delay or misdiagnosis in women. These inequalities directly affect health equity, especially in medical deserts, where marginalized communities already face limited access to care.

Previous research has explored AI’s clinical applications, including diagnostics, predictive analysis, and patient monitoring (Ahmed et al., 2023; Bajwa et al., 2021), as well as challenges such as bias and regulatory concerns (Liu et al., 2025; Olawade et al., 2024). However, existing literature does not always go into detail about the integration of AI into the clinical and administrative sides of healthcare. Specifically, research rarely examines how AI integration affects workflow efficiency, administrative burden, patient engagement, and the role of clinician oversight. The research question guiding this paper is: “In what ways is artificial intelligence reshaping clinical care, administration, and organizational practices in Pittsburgh’s healthcare sector?” This paper hopes to understand operational and organizational impacts from the implementation of AI into healthcare in Pittsburgh.

Synthesis

AI in healthcare has had a lot of areas of growth and development since it is an important area of impact. An example of this is a paper that came out in 1956 titled “Summer Research Project of Artificial Intelligence” (Ahmed et al., 2023). This article talked about the initial exploration of AI’s role in medicine and how it will develop in the future. Many current articles examining AI’s clinical application also talk about the importance of early integration of AI into healthcare that is based on expert systems and decision support tools (Olawade et al., 2024). AI can become more reliable with early integration since it can also lead to early training and understanding how AI can efficiently integrate into healthcare to support the patients and clinicians. AI is also growing through the adoption of technology in medicine over the past 50 years, especially through the evolution, introduction, and integration of machine learning, deep learning, and large language models. An example of this spoken about in the literature was the innovation of diagnosing breast cancer through a system which integrated AI; it was one of the first cases of using AI in healthcare. This led to an increased focus on personalized medicine, diagnostics, and predictive analysis (Ahmed et al., 2023). These tools are beneficial to healthcare, especially with precision and improving the quality of care; however, some negative effects could be improved upon with more human involvement.

AI improves precision and quality of healthcare due to its accurate diagnosis of various diseases, like diagnosing breast cancer; this frees clinicians to focus on patient engagement. The use of AI can lead to more personalized treatments based on various factors in a patient’s medical history or genetics, which can be easily accessed through AI. However, integrating AI with human involvement will also aid in improving the quality of healthcare, as humans are necessary to review the results and make final treatment decisions. There are also many current applications of AI in healthcare, including AI chatbots; imaging in fields such as dermatology, radiology, and pathology, precision therapeutics; and CRISPR gene editing (Bajwa et al., 2021). For example, AI has been applied to genomic surveillance in outbreak investigations, identifying transmission routes more efficiently and accurately than manual review (PubMed, 2023). Similarly, UPMC has implemented “digital twinning” technology, allowing clinicians to model individual patients’ demographics, health conditions, and medications to simulate potential treatment outcomes (Katz, 2023). However, these applications are being further improved upon with the co-innovation of AI and clinical organizations. This combination of structured and unstructured data from both AI and human healthcare professionals, creates a human-centered AI (Bajwa et al., 2021). A positive change of this modeling is that the quality of healthcare is being improved by AI, while having human integration makes a more accurate clinical diagnosis and increases trust that decisions made about patient care are ethical.

There are many benefits of AI in healthcare as well, since it has many applications that help with facilitating early diagnosis, monitoring health parameters, and increasing the quality of care. AI can facilitate procedures that have previously been performed, becoming a repetition and prediction model, leading to increased free clinical time for patient-clinician interactions and streamlining care pathways (Bajwa et al., 2021; Ahmed et al., 2023). An example of this is AI assistance in early disease detection, where AI can pre-scan images instead of having radiologists examine each image in detail. This use of AI, in return, allows physicians to have more time in discussing the best treatment plan for each patient. Clinicians also experience reduced administrative burden and improved workflow with AI integration, enhancing their allocation of their time and resources. For example AI powered medical EHR systems, such as Epic, assist in transcribing and organizing patient conversations into medical documentation by assisting in drafting responses to questions. Not only that, but medical deserts are further improved upon with AI-driven analytics and optimized staffing, which can identify healthcare provision gaps, and therefore bridge healthcare gaps to improve access and quality (Zdeslav Strika et al., 2024).

Although there are many concerns with AI bias and AI fairness, AI can play a critical role in promoting equitable care by supporting bias detection and mitigation. AI can also make sure that there is more equitable allocation of healthcare resources, including clinical time, diagnostic services, and treatment access. AI assists in detecting bias in healthcare as it can analyze datasets to identify patterns of unequal treatments, and can use results to assist in bias mitigation. Not only that, it can also help with allocating limited resources based on health data which is more accurate. There are many underrepresented populations in datasets, however, and this makes it so that AI is unable to recognize patterns on a wider dataset, leading to AI bias (Liu et al., 2025). The most important fixes to this is to gather data that includes a broader representative set of the population and patients, and developing methods that detect and mitigate bias during training, showing that with more human involvement in AI data analysis, there will be fewer risks of bias.

There are also many barriers and negative effects of AI in healthcare, especially in sectors such as privacy, consent, and conflict of interest. There has also been a lot of debate over patient data ownership and protecting HIPAA regulations and confidentiality with the integration of AI to analyze vast sets of data (Liu et al., 2025). AI is also able to easily integrate into healthcare’s existing data quality and organizational workflow, but there are still multiple debates surrounding the ability of AI tools and the overestimation of its capabilities (Olawade et al., 2024). This is especially due to the inability of AI to function optimally in multiple sectors of healthcare without human involvement. Two of the largest barriers to AI’s integration into healthcare include the lack of consistent regulations across healthcare systems and the clinical responsibility of AI to make decisions (Zdeslav Strika et al., 2024). With the integration of AI in healthcare, there must be many regulations and constraints since, essentially, a mathematical model will carry a lot of responsibilities related to the direct outcome of patient care. There are many contrasting perspectives on the need for regulation and how it will slow innovation; however, a phased approach could help. Taking different incremental steps toward the integration of AI into healthcare would still require consistent regulations across healthcare systems.

Implementing AI in healthcare also increases the need for training and education and comes with a fear of job displacement (Ahmed et al., 2023). However, with consistent regulation, there will also be a lot more access to educational programs. An example is how the European Commission’s DigComp framework key competencies that are necessary to adapt new technologies, especially integration of AI into healthcare.

Many studies have also shown that training healthcare professionals about the integration of AI into healthcare may also require specialized roles and professionals to ease the adaptation and use of AI in clinical settings. (Gazquez-Garcia et al., 2025). Not only that, but in a sector with deep human connections and involvement like healthcare, the optimal use of AI requires human-AI collaboration. There are also many concerns for patient safety, as industry leaders debate the reliability of AI use even with human oversight (Olawade et al., 2024). One of these concerns is AI misdiagnosis, especially when misdiagnosis can lead to the wrong treatment plan and delayed care. Case studies from UPMC Children’s Hospital illustrate a deliberate, multidisciplinary approach to AI integration, with projects like Ambient AI being trialed for clinical documentation and AI-driven sepsis detection showing promise, which ensures clinician buy-in and patient safety (MobiHealthNews, 2023). Similarly, Dr. Andrew Watson’s work on telehealth and AI-assisted post-acute monitoring emphasizes humane, patient-centered approaches. Watson’s AI assisted remote monitoring for patients post-surgery leverages digital tools to improve efficiency and reduce travel-related risks (YouTube, 2023). However, with consistent regulation that will be able to go over the boundaries and barriers stated by AI, the implementation of AI can positively benefit both the caretakers and patients in healthcare.

It is important to address bias through clinical involvement and multidisciplinary collaboration, especially with human-centered AI approaches to design, experimentation, and evaluation of the sources. Through consistent regulations and leveraging AI to bridge healthcare gaps in underserved communities, AI will be beneficial to healthcare through telemedicine and early diagnosis and screening to improve outcomes in areas with limited medical staff. (Olawade et al., 2024). AI has vast potential to address systemic barriers, and AI adoption is required to balance innovation with ethics, fairness, and improved workforce readiness (Olawade et al., 2024).

Methodology

The main method that will be used to determine the effects of the implementation of AI into healthcare, specifically on the administrative and operational side, is case studies. This study uses a case study to explore how AI is integrated into the healthcare system in Pittsburgh, specifically in UPMC, and its impact on clinical and administrative practices. The two primary case studies that are used to perform this research are “UPMC Children’s Hospital AI Projects” and “Operationalizing Digital Health at UPMC,” both conducted in 2023. In UPMC Children’s Hospital AI Projects byMobiHealthNews, there were many topics that were spoken about, most importantly, how AI tools are implemented slowly and responsibly to meet real clinical needs. A precaution taken in these projects is making sure that there is human oversight while implementing the AI gradually. It also talks about the Ambient AI clinical documentation program, which is already implemented in certain departments and saves time for patient care. Specifically, it focuses on using microphones to generate clinical notes for physicians during clinical interactions. AI is also being applied in other ways, including sepsis detection, diabetes management, and autism care.

However, a documentary about these trials available on YouTube, Operationalizing Digital Health, also talks about how AI integration into healthcare requires multidisciplinary teams, which includes clinicians, regulators, and IT staff. This is important as it shows how the integration of AI into healthcare has multiple teams assisting it, ensuring that AI tools are safe and effective, while also being relevant.

The second case study that was looked into in Operationalizing Digital Health was from 2023. This interview discussed how AI supports telemedicine, post-acute monitoring, and focused workflow solutions. In the case study, Dr. Andrew Watson spoke about how digital tools save a lot of time and improve care accessibility, and AI integration requires attention to workflow optimization. This ensures that AI tools automate repetitive tasks, and fit efficiently into clinical processes to save time and improve access to quality healthcare. The independent variable in this study is the AI integration strategy, which is a structured, phased implementation with limited integration; the dependent variable is the clinical efficiency and patient care outcomes. Data from the cases were analyzed qualitatively by using patterns in operational efficiency and clinical experiences.

Results

The UPMC case studies reveal that AI integration into healthcare, when it is more structured and has human integration as well, leads to measurable improvements in clinical and operational outcomes. An example of this is that Ambient AI documentation saves clinician time for increasing patient interactions while also increasing efficiency. Maximizing clinician time for  patient care improves the overall experience for patients and healthcare efficiency as there is more face to face interaction allowing clinicians to thoroughly assess the patient and personalize the treatment (MobiHealthNews, 2023). Some other examples of the integration of AI into healthcare from research papers include digital twinning and AI-assisted monitoring to improve personalized care and treatment planning for patients (Katz, 2023).  As discussed above, “digital twinning” and AI assisted monitoring makes the model of healthcare more personalized to each patient, as it allows physicians to use a virtual replica of a patient. Not only that, but AI genomic surveillance tools also enhance investigations of outbreaks of different illnesses and viruses with 91.7% accuracy compared to other reviews (PubMed, 2023). This accuracy finding is important as it shows how AI increases the efficiency of many healthcare procedures. However, it is also important that there is human integration as well, which  can be done by phased integration of any AI tool, which these case studies suggest is critical for clinician buy-in and the ultimate adoption of AI in clinical practice.

Discussion

These findings indicate that AI can enhance healthcare delivery beyond clinical accuracy, improving workflow efficiency, patient engagement, and safety when integrated responsibly. AI improves workflow since it assists with automating repetitive tasks allowing clinicians to focus on patient care, especially when it comes to documentation and charting, reducing clinical burnout and increasing the number of patients seen. Not only that, AI also improves patient engagement as it allows clinicals to provide more personalized and accessible care. Patient safety is also improved as AI assists in detecting errors earlier making sure that errors are prevented and bias is detected and mitigated, supporting prior research on AI’s potential to reduce administrative burden (Ahmed et al., 2023; Bajwa et al., 2021) while extending it to organizational and operational contexts. The research highlights that human-centered AI and multidisciplinary collaboration are essential to mitigating risks like bias, ethical violations, or workflow disruption (Liu et al., 2025; Olawade et al., 2024). Findings also complement Katz’s (2023) insights on digital twinning and PubMed's (2023) genomic surveillance, illustrating that AI can serve both clinical and operational needs simultaneously. This is done through AI’s assistance in diagnosing, creating treatment plans, and monitoring patients. Not only that, on a more operational level it assists in optimizing hospital efficiency and resource allocation. However, there are some limitations as well. Case studies focus on only UPMC, limiting generalizability, and we do not understand the long-term impact on patient outcomes since it requires longitudinal studies. However, AI can reshape healthcare administration and clinical workflows in meaningful ways, but success depends on structured implementation, clinician engagement, and ethical oversight. Future research should expand to multiple institutions, assess long-term impacts, and explore AI’s role in reducing health disparities, especially in medical deserts.

References