Artificial Intelligence in Healthcare

 

Artificial Intelligence (AI) technologies have the potential to make many positive changes in the field of healthcare.  AI can have an impact on the provider side, the customer/patient side, and the administrative side of healthcare.  For some examples, let’s examine the following applications:

1.      Medical Imaging and Diagnostics

2.      Virtual Health Assistants

3.      Data Management

Medical Imaging and Diagnostics

                Medical Imaging is one area where AI may be able to assist healthcare providers in allowing for faster and more accurate diagnostics.  One example of AI-assisted imaging is the system designed by Imperial College London and Edinburgh University.  This system can detect strokes, determine when the stroke happened, and what treatment would be appropriate.  This system was built on a Machine Learning model that was trained with the brain scans from 783 patients who had suffered a stroke, whose symptoms and approximate time of the stroke were known to the medical providers treating the patients.  The system was tested with 1,945 patients, and results indicate that the AI-enhanced system was approximately twice as accurate in detecting strokes in patients when compared to current methods of detection (Marcus et al., 2024).

                The diagnosis and prediction of illness is also an area where AI may assist, as it can examine vast quantities of health data.  One interesting example is the Machine Learning with Phenotype Associations (MILTON).  This system was trained with genetic data obtained from the UK Biobank, comprising nearly 500000 individuals.  This system may be able to detect early signs of illnesses that might not become apparent until years later.  MILTON is able to detect these diseases earlier than current methods by examining patients’ genomic data for certain biomarkers, and in fact, it can outpace current methods of detection for the same illnesses such as diabetes and Alzheimer’s (Garg et al., 2024).

Virtual Health Assistants

                The public is already using LLMs, such as ChatGPT, Gemini, Claude, etc.), to ask for health information.  A study conducted by Low et al., n.d., shows that the accuracy of LLMs for health-related questions may be around 2-10%.  The researchers suggest linking an out-of-the-box LLM to a Retrieval Augmented Generation (RAG) agent that will retrieve information from curated sources based on real-world data and experience.  When answers were analyzed, it showed that the accuracy of responses was between 24-58%.  Needless to say, these systems are still not perfect, but there is a space where AI-enhanced virtual health assistant are going to be able to make an impact in patient healthcare.  As technology improves and the accuracy of systems that are guided by health professionals rather than generic algorithms increases, patients may feel comfortable with the adoption of virtual health assistants.

Data Management

                This is probably the area in healthcare where AI technology might be more easily adopted and implemented.  AI has the ability to examine vast amounts of data rapidly, freeing up human operators from mundane and repetitive tasks.  This frees up different health professionals to focus on critical tasks rather than administrative tasks.  One challenging area for sharing electronic health records (EHR) is system interoperability; AI-enhanced systems may be able to assist in the exchange of EHR between separate health information systems that normally face difficulties due to inadequate interoperability (Orsi, 2024).  Another useful AI-enhanced system is Amazon HealthScribe, which can convert conversations that medical providers have with patients into digitized medical notes that can be quickly transferred into the patient’s EHR, instead of having to manually transcribe discussions that providers have with their patients (Amazon Web Services, 2026).

Ethical Considerations of AI in Healthcare

                The major ethical consideration in AI applications in healthcare is the use of data without appropriate consent.  AI models require vast amounts of data in order to function as accurately as possible, and accuracy makes a tangible impact on patient health outcomes.  This brings up an interesting dilemma; as healthcare consumers, we want providers to have the best knowledge to diagnose and provide treatments, but at the same time, we want our private health information to be protected or at least not used without our consent.  Without access to patient data, AI models may not be appropriately trained, and if AI models are not appropriately trained, medical errors may emerge which may result in patient harm (Farhud & Zokaei, 2021).  Current health information protection regulations such as HIPAA should be updated to include AI technologies, and how patient health information may be used for AI models.  Healthcare organizations should be required to inform their patients that their health information may be used for AI training purposes, how it is going to be used, what safeguards are going to be used, and for how long the patient information is going to be maintained by the AI operators (Horton & Lucassen, 2022).

Future Prospects for AI in Healthcare

                Although AI technologies are currently not accurate and responsive to the point where the general public feels completely comfortable with their adoption, the future of healthcare will be intertwined with AI technologies.  AI will be implemented in healthcare systems to automate routine tasks that can take time away from direct patient care; I believe that would be the most appropriate function for AI at this point in the development of the technology.  AI-enhanced data management has the potential to be a game changer, and I am confident that it will have a strong impact in the field of health information systems.  AI-enhanced medical research, and AI-assisted medical diagnosis are probably going to be implemented next as more accuracy is built into the systems and as medical providers and researchers become more familiar with these technologies.  Finally, AI-enhanced patient interfaces might be the one area that will face the most challenges, as people might not be fully confident in how these systems function, their accuracy, and the lack of real empathy that can only be found in real human contact.  As healthcare organizations look for ways to reduce costs, AI patient assistants will be implemented; how they will be received by the public can’t be predicted; it could be a success, but it could end up being a complete failure.

 

 

References:

Amazon Web Services. (2026, April 9). What is AI in Healthcare? Amazon Web Services, Inc. https://aws.amazon.com/what-is/ai-in-healthcare/#what-are-the-applications-of-ai-in-healthcare--1mdtv0p

Farhud, D., & Zokaei, S. (2021). Ethical issues of artificial intelligence in medicine and healthcare. Iranian Journal of Public Health, 50(11), 1–5. https://doi.org/10.18502/ijph.v50i11.7600

Garg, M., Karpinski, M., Dorota Matelska, Middleton, L., Burren, O. S., Hu, F., Wheeler, E., Smith, K. R., Fabre, M. A., Mitchell, J., O’Neill, A., Ashley, E. A., Harper, A. R., Wang, Q., Dhindsa, R. S., Petrovski, S., & Dimitrios Vitsios. (2024). Disease prediction with multi-omics and biomarkers empowers case–control genetic discoveries in the UK Biobank. Nature Genetics, 56(9), 1821–1831. https://doi.org/10.1038/s41588-024-01898-1

Horton, R., & Lucassen, A. (2022). Ethical Considerations in Research with Genomic Data. The New Bioethics, 29(1), 1–15. https://doi.org/10.1080/20502877.2022.2060590

‌Low, Y., Jackson, M., Hyde, R., Brown, R., Sanghavi, N., Baldwin, J., Pike, C., Muralidharan, J., Hui, G., Alexander, N., Hassan, H., Nene, R., Pike, M., Pokrzywa, C., Vedak, S., Yan, A., Yao, D.-H., Zipursky, A., Dinh, C., & Ballentine, P. (n.d.). Answering real-world clinical questions using large language model based systems. https://arxiv.org/pdf/2407.00541

Marcus, A., Mair, G., Chen, L., Hallett, C., Cuervas-Mons, C. G., Roi, D., Rueckert, D., & Bentley, P. (2024). Deep learning biomarker of chronometric and biological ischemic stroke lesion age from unenhanced CT. Npj Digital Medicine, 7(1). https://doi.org/10.1038/s41746-024-01325-z

Orsi, E. (2024, April 26). How AI Is a Game Changer for Healthcare Data Management. Laserfiche. https://www.laserfiche.com/resources/blog/how-ai-is-a-game-changer-for-healthcare-data-management/

Comments

Post a Comment

Popular posts from this blog

  Searching for Health Information Online, a Survey. People interested in health information have been using the Internet to search ever since it became available to the general public.   Per the Pew Research Center (2005), at least 80% of Internet users had looked for health information online by the year 2002.   That number is nearly identical to a study conducted in 2018, where 79.12% of respondents said they had used the Internet to look for health information for themselves within the previous twelve months.   Additionally, 41.01% of respondents had used the Internet for healthcare decisions, and 39.37% said they had used it for discussions with healthcare providers (Taya & Chuang, 2025).   These facts lead to the questions: “How do people look for online health information?” “How do they evaluate the health information that they view online?” and finally, “how comfortable are they with AI applications providing health information?” Methodology The ...
Read more