ACE

Tag: Hospitals

  • Understanding the Debate on AI in Electronic Health Records

    Understanding the Debate on AI in Electronic Health Records

    Background

    Artificial Intelligence (AI) refers to the use of computer algorithms to process data and make decisions, ultimately streamlining administrative processes. In healthcare, AI is being increasingly integrated with Electronic Health Records (EHRs)—digital systems that store and manage patient health information, such as medical history and diagnoses. By 2021, almost 80% of office-based physicians and virtually all non-federal acute care hospitals had implemented an EHR system. As part of this widespread adoption, various AI applications in EHRs are beginning to emerge. So far, the main functions of AI in EHRs include managing datasets of patient health information, identifying patterns in health data, and using these patterns to predict health outcomes and recommend pathways for treatment. 

    Arguments in Favor of AI in EHRs

    The use of AI in EHRs presents opportunities to improve healthcare by increasing efficiency as well as supplying administrative support. Supporters of AI integration argue that it can significantly improve diagnostic accuracy. AI-integrated EHR systems can analyze vast amounts of patient data, flagging potential issues that might otherwise be overlooked by human clinicians. Machine learning algorithms can identify patterns across multiple cases and recommend diagnoses or treatments based on evidence from similar cases. Proponents contend that by reducing human error and providing real-time insights, AI could support doctors in making more accurate and quick decisions, leading to better patient outcomes.

    Proponents of AI in EHRs also argue that AI has the potential to significantly reduce healthcare inequities by providing better access and more personalized care for underserved populations. AI-powered tools can identify at-risk patients early by analyzing complex data, including demographic and behavioral factors, and help prioritize interventions for those who need it most. Additionally, AI can bridge communication gaps for patients facing language barriers or low health literacy, ensuring they receive clear and relevant information about their health. Supporters also suggest that AI’s ability to reduce human biases in clinical decision-making, such as disparities in pain assessment or treatment recommendations, could lead to fairer, more equitable healthcare outcomes for all.

    From the workforce perspective, supporters argue that AI integration in EHRs has the ability to significantly reduce physician burnout by streamlining the documentation process. With the increasing time spent on EHR tasks, AI-driven tools like voice-to-text transcription, automated note generation, and data entry can cut down the time physicians devote to administrative duties. For instance, one 2023 study reported that AI integration in health records led to a 72% reduction in documentation time, equating to approximately 3.3 hours saved per week per clinician. This allows doctors to spend more time on direct patient care and less on paperwork, which supporters contend will improve job satisfaction and reduce stress.

    Arguments Against AI in EHRs

    While some argue that AI in EHRs will lead to more accurate and equitable healthcare, others raise concerns regarding data bias, privacy, and transparency. Critics of AI integration argue that modern legal frameworks lack adequate safeguards for individuals’ health data, leaving sensitive information vulnerable to breaches. For example, data collected by AI tools may be hacked or gathered without consent for marketing purposes. Additionally, certain genetics testing companies that operate without sufficient legal oversight may sell customer data to pharmaceutical and biotechnology companies.

    Moreover, some critics share concerns about whether AI integration in EHRs aligns with standards for informed consent. Informed consent is a key ethical principle that ensures patients are fully informed and in control of decisions regarding their healthcare. It includes elements such as the patient’s ability to understand and make decisions about their diagnoses, treatment options, and any risks involved. Ethical responsibility dictates that consent should be specific, voluntary, and clear. The rise of AI in healthcare applications has increased concerns about whether patients are fully aware of how their data is used, the risks of procedures, and potential errors in AI-driven treatments. Autonomy principles state that patients have the right to be informed about their treatment process, the privacy of their data, and the potential risks of AI-related procedures, such as errors in programming. Critics say that patients must be more informed about how AI is integrated into health records systems in order for them to truly provide informed consent. 

    Another significant ethical concern in the use of AI and machine learning (ML) in healthcare is algorithmic bias, which can manifest in racial, gender, and socioeconomic disparities due to flaws in algorithm design. Such biases may lead to misdiagnosis or delayed treatments for underrepresented groups and exacerbate inequities in access to care. To address this, advocates push for the prioritization of diverse training data that reflects demographic factors. They hold that regular evaluations are necessary to ensure that AI models consistently remain fair over time, upholding the principles of justice and equity. 

    Future Outlook

    Building on the potential of AI in healthcare, H.R. 238, introduced on January 7, 2025, proposes that AI systems be authorized to prescribe medications if they are approved by the Food and Drug Administration (FDA) and if the state where they operate permits their use for prescribing. This bill represents a significant step in integrating AI into clinical practices, going beyond data management to reshape how medications are prescribed and managed. The arguments for and against H.R. 238 mirror the debate around AI integration in EHRs; while proponents of the bill argue that AI could enhance patient safety, reduce errors, and alleviate clinician burnout, critics highlight concerns regarding the loss of human judgment, data privacy, and the potential for AI to reinforce biases in healthcare. As AI continues to play a central role in healthcare, bills like H.R. 238 spark important discussions about AI’s ethical, practical, and legal implications in clinical decision-making.

    Summary

    In conclusion, the integration of AI into EHRs has forced medical stakeholders to balance a need for improvements in accuracy and efficiency with a concern for medical ethics and patient privacy. On one hand, AI can support more accurate diagnoses, enhance patient care, and help reduce the burnout faced by healthcare providers. Additionally, AI may contribute to reducing healthcare inequities by providing better access and more personalized care, especially for underserved populations. However, the implementation of AI also raises concerns regarding data privacy, algorithmic bias, and informed consent, suggesting a need for more careful implementation and oversight. As AI’s presence in healthcare settings continues to expand, addressing these concerns will be key to ensuring it benefits patients and healthcare providers alike.

  • Understanding the AI in Healthcare Debate

    Understanding the AI in Healthcare Debate

    Background

    What is Artificial Intelligence?

    Artificial intelligence, more commonly referred to as AI, encompasses many technologies that enable computers to simulate human intelligence and problem solving abilities. AI includes machine learning, which allows computers to imitate human learning, and deep learning, a subset of machine learning that simulates the decision making processes of the human brain. Together, these algorithms power most of the AI in our daily lives, such as Chat GPT, self-driving vehicles, GPS, and more. 

    Introduction

    Due to the rapid and successful development of AI technology, its use is growing across many sectors including healthcare. According to a recent Morgan Stanley report, 94 percent of surveyed healthcare companies use AI in some capacity. In addition, a MarketsandMarkets study valued the global AI healthcare market at $20.9 billion for 2024 and predicted the value to surpass $148 billion by 2029. The high projected value of AI can be attributed to the increasing use of AI across hospitals, medical research, and medical companies. Hospitals currently use AI to predict disease risk in patients, summarize symptoms for potential diagnoses, power chatbots, and streamline patient check-ins. 

    The increased use of AI in healthcare and other sectors has prompted policymakers to recommend global standards for AI implementation. UNESCO published the first global standards for AI ethics in November 2021, and the Biden-Harris Administration announced an executive order in October 2023 on safe AI use and development. Following these recommendations, the Department of Health and Human Services published a regulation titled HTI-1 Final Rule, which includes requirements, standards, and certifications for AI use in healthcare settings. The FDA also expanded its inspection of medical devices that incorporate AI in 2023, approving 692 AI devices. While the current applications of AI in the health industry seem promising, the debate over the extent of its use remains a contentious topic for patients and providers.

    Arguments in Favor of AI In Healthcare

    Those in favor of AI in healthcare cite its usefulness in diagnosing patients and streamlining patient interactions with the healthcare system. They point to evidence showing that AI is valuable for identifying patterns in complex health data to profile diseases. In a study evaluating the diagnostic accuracy of AI in primary care for over 100,000 patients, researchers found an overall 84.2 percent agreement rate between the physician and the AI diagnosis. 

    In addition, proponents argue that AI will reduce the work burden on physicians and administrators. According to a survey by the American Medical Association, two thirds of over 1,000 physicians surveyed identified advantages to using AI such as reductions in documentation time. Moreover, a study published in Health Informatics found that using AI to generate draft replies to patient messages reduced burnout and burden scores for physicians. Supporters claim that AI can improve the patient experience as well, reducing waiting times for appointments and assisting in appointment scheduling.

    Proponents also argue that using AI could significantly combat mounting medical and health insurance costs. According to a 2024 poll, around half of surveyed U.S. adults said they struggled to afford healthcare, and one in four said they put off necessary care due to the cost. Supporters hold that AI may be a solution, citing one study that found that AI’s efficiency in diagnosis and treatment lowered healthcare costs compared to traditional methods. Moreover, researchers estimate that the expansion of AI in healthcare could lead to savings of up to $360 billion in domestic healthcare spending. For example, AI could be used to save $150 billion annually by automating about 45 percent of administrative tasks and $200 billion in insurance payouts by detecting fraud. 

    Arguments Against AI in Healthcare

    Opponents caution against scaling up AI’s role in healthcare because of the risks associated with algorithmic bias and data privacy. Algorithmic bias, or discriminatory practices taken up by AI from unrepresentative data, is a well-known flaw that critics say is too risky to integrate into already-inequitable healthcare settings. For example, when trained with existing healthcare data such as medical records, AI algorithms tended to incorrectly evaluate health needs and disease risks in Black patients compared to White patients. One study argues that this bias in AI medical applications will worsen existing health inequities by underestimating care needs in populations of color. For example, the study found that an AI system designed to predict breast cancer risk may incorrectly assign Black patients as “low risk”. Since clinical trial data in the U.S. still severely underrepresents people of color, critics argue that algorithmic bias will remain a dangerous feature of healthcare AI systems in the future.

    Those against AI use in healthcare also cite concerns with data privacy and consumer trust. They highlight that as AI use expands, more corporations, clinics, and public bodies will have access to medical records. One review explained that recent partnerships between healthcare settings and private AI corporations has resulted in concerns about the control and use of patient data. Moreover, opponents argue that the general public is significantly less likely to trust private tech companies with their health data than physicians, which may lead to distrust of healthcare settings that partner with tech companies to integrate AI. Another issue critics emphasize is the risk of data breaches. Even when patient data is anonymized, new algorithms are capable of re-identifying patients. If data security is left to private AI companies that may not have experience protecting such large quantities of patient data against sophisticated attacks, opponents claim the risk of large-scale data leaks may increase. 

    Conclusion

    The rise of AI in healthcare has prompted debates on diverse topics ranging from healthcare costs to work burden to data privacy. Proponents highlight AI’s potential to enhance diagnostic accuracy, reduce administrative burdens on healthcare professionals, and lower costs. Conversely, opponents express concerns about algorithmic bias exacerbating health disparities and data breaches leaking patient information. As the debate continues, the future of AI in healthcare will hinge on addressing these diverse perspectives and ensuring that the technology is developed responsibly.