By - Pooja Suganthan, Edited By - Melissa-Maria Kulaprathazhe
There was a story of a 52-year old woman who was admitted to an emergency room in Baltimore due to a sore foot. The emergency medical services diagnosed her and transferred her to the general ward. On her third day in the hospital, the woman showed signs of pneumonia and was given appropriate antibiotics. However, three days later, her heartbeat and breathing rate rose rapidly. Her condition worsened in the following twelve hours. She entered septic shock (when blood pressures fall and organs do not receive enough oxygen)- a condition with a mortality rate as high as 50%. For the next three weeks, the patient was moved to the ICU and given proper care and treatment, but it was unfortunately too late. The woman passed away on her twenty-second day in the hospital (Ashley, 2017).
This patient’s fate, like many others with sepsis, was not due to the lack of treatment, but rather a delay in detection. A promising solution to this problem is the implementation of artificial intelligence systems in healthcare. According to Suchi Saria, the director of the Machine Learning lab at Johns Hopkins University, “The signs that help pinpoint the diagnosis may already be in your data, but the data in electronic medical records are really messy.” Saria’s team is working on “designing computer algorithms based on statistical and computational techniques that were recently developed to allow clinical experts to identify sepsis faster” (TEDx Talks, 2016).
Artificial intelligence (AI) are software systems that use algorithms to make decisions and respond to stimuli in real time. These algorithms combine a wide range of information obtained from sensors to digital data. Then, they input data from pre-existing databases. The AI software analyzes the combined information to curate an appropriate response (West, 2018). So, researchers, software engineers, and healthcare professionals are looking at ways to incorporate artificial intelligence technologies in healthcare. Artificial intelligence consists of several types of technologies. Some types of AI that are relevant to healthcare include machine learning, natural language processing, rule-based expert systems, physical robots, and robot process automation. Machine learning is one of the more popular techniques. It involves systems that learn from provided data and learn from experience, rather than being programmed to behave in a fixed and rigid manner. An application of machine learning would be an algorithm that can predict successful treatment protocols based on a patient’s symptoms and pre-existing datasets (Davenport & Kalakota, 2019).
There are several key advantages of implementing artificial intelligence in hospitals across the nation. AI systems can aid in diagnosis and treatment because they can be trained to detect certain symptoms. These algorithms are commonly used by radiologists for cancer detection. A study of a breast cancer screening AI system showed that the algorithm’s cancer detection accuracy was comparable to that of an average radiologist (Rodriguez-Ruiz et al., 2019). Since the machines would be using data from past patients to make their diagnosis, there is potential for earlier diagnoses. Research conducted by Google AI shows how three-dimensional deep learning systems can screen for lung cancer faster than and with higher accuracy than trained radiologists (Ardila et al., 2019). This is advantageous because “doctors have a much longer window to come in and intervene in order to prevent organ dysfunction and mortality” (TEDx Talks, 2016). Some worry that incorporating artificial intelligence into healthcare would take away jobs and make healthcare feel more automated but artificial intelligence systems can help alleviate some tasks from workers. However, these systems typically perform one specific task. For example, AI systems in radiology can diagnose and group images, but they still require radiologists to perform other responsibilities, such as consulting with other professionals, providing treatment, and drawing conclusions from the AI provided data (Davenport & Kalakota, 2019).
Thus, AI systems should not be thought of as a replacement for healthcare professionals, but rather as a supplemental tool. However, there are some ethical concerns that need to be addressed before the widespread implementation of AI systems. For example, AI systems pose “threats to privacy and confidentiality, informed consent, and patient autonomy—and to consider how AI is to be integrated in clinical practice” (Rigby, 2019). Another potential issue is unconscious bias. An AI system’s decision is influenced by the type of data it was trained with. So, if there is a particular demographic with inherently biased data or inadequate representation, the AI’s output will reflect this and therefore not be accurate (Artificial Intelligence Bias in Healthcare, n.d.). In addition, healthcare professionals need to learn how to incorporate these new technologies into their practice. One suggestion would be to reframe medical schools’ curricula to incorporate these potential tools (Rigby, 2019).
In summary, artificial intelligence systems can revolutionize healthcare. They can assist professionals with patient care, administrative tasks, diagnosis, and treatment. However, these systems cannot substitute for a healthcare professional because AI algorithms present unconscious bias. Researchers, engineers, and healthcare professionals need to work together to find a safe, ethical, accurate balance.
Reference(s)
Ardila, D., Kiraly, A. P., Bharadwaj, S., Choi, B., Reicher, J. J., Peng, L., Tse, D., Etemadi, M., Ye, W., Corrado, G., Naidich, D. P., & Shetty, S. (2019). End-to-end lung cancer screening with three-dimensional deep learning on low-dose chest computed tomography. Nature Medicine, 25(6), 954–961. https://doi.org/10.1038/s41591-019-0447-x Artificial Intelligence Bias in Healthcare. (n.d.). Retrieved October 21, 2020, from https://www.boozallen.com/c/insight/blog/ai-bias-in-healthcare.html Ashley, S. (2017, October 11). Using Artificial Intelligence to Spot Hospitals’ Silent Killer. https://www.pbs.org/wgbh/nova/article/ai-sepsis-detection/ Davenport, T., & Kalakota, R. (2019). The potential for artificial intelligence in healthcare. Future Healthcare Journal, 6(2), 94–98. https://doi.org/10.7861/futurehosp.6-2-94 Rigby, M. J. (2019). Ethical Dimensions of Using Artificial Intelligence in Health Care. AMA Journal of Ethics, 21(2), 121–124. https://doi.org/10.1001/amajethics.2019.121. Rodriguez-Ruiz, A., Lång, K., Gubern-Merida, A., Broeders, M., Gennaro, G., Clauser, P., Helbich, T. H., Chevalier, M., Tan, T., Mertelmeier, T., Wallis, M. G., Andersson, I., Zackrisson, S., Mann, R. M., & Sechopoulos, I. (2019). Stand-Alone Artificial Intelligence for Breast Cancer Detection in Mammography: Comparison With 101 Radiologists. JNCI: Journal of the National Cancer Institute, 111(9), 916–922. https://doi.org/10.1093/jnci/djy222 TEDx Talks. (2016, October 12). Better Medicine Through Machine Learning | Suchi Saria | TEDxBoston. https://www.youtube.com/watch?v=Nj2YSLPn6OY West, D. M. (2018, October 4). What is artificial intelligence? Brookings. https://www.brookings.edu/research/what-is-artificial-intelligence/