In recent years, U.S. physicians have been reporting high levels of burnout, stress, and exhaustion. This is partly due to the demands of the global COVID-19 pandemic coupled with the stress of navigating what many believe to be a broken healthcare system (Definitive Healthcare, 2022). In addition, the Association of American Medical Colleges reports that within the next twelve years, the United States will face a shortage of between 37,800 and 124,000 physicians (Robeznieks et al., 2022). A source of relief for the overworked and failing parts of the healthcare system may lie in the potential of artificial intelligence (AI).
AI has the potential to revolutionize the medical field by automating diagnostic procedures and easing physicians’ workloads (Sharif et al., 2023). Scientists have successfully designed an AI model that can predict future risk of cardiac incidents using spinal imaging. Bone density machines are used to capture the spinal images that are then assigned an abdominal aortic calcification (AAC) score (Sharif et al., 2023). AACs occur when calcium crystals develop in the abdominal aorta, a large blood vessel that runs from the diaphragm to the pelvis (Patients Like Me, 2023). Recent meta-analyses, which combine the results of multiple scientific studies, have shown an association between AAC scores and the risk of future cardiovascular incidents and mortality (Patients Like Me, 2023). Typically, trained imaging specialists assign AAC scores manually, which is inefficient and labor-intensive. Using a collection of already scored spinal images, researchers sought to train a machine learning algorithm to be able to assign scores to the images. They then tested the algorithm and compared the scores assigned by AI to real-world cardiovascular outcomes (Sharif et al., 2023). The machine learning algorithm’s score matched humans’ score about 80 percent of the time and was associated with a high gradient of risk for cardiac events, which means the higher scores were related to undue pressure in the aortic valve (Pick, 2019). The success rate is impressively high, especially considering that this is the first official version of the algorithm. Additionally, the machine is much more efficient compared to its human counterparts and can process around 60,000 images a day and mistakenly assigns a reduced AAC only 3% of the time (Sharif et al., 2023). These models’ predictions will allow patients to be more aware of their health status and modify their habits accordingly if they are at risk of a cardiac event.
This study, though one of the most far-ranging, is not the only of its kind. Artificial intelligence models are showing great promise throughout the field of disease detection and diagnosis. For instance, Janghel and his team used Magnetic Resonance Images of those with Alzheimer’s and those of healthy individuals for a deep learning technique program. The program had a 99.95% accuracy rate for correctly identifying patients with Alzheimer’s. Since there is not yet an approved and effective cure for Alzheimer's disease, physicians rely on medications to delay the progress of the illness by making early detection (Janghel et al., 2020). Early detection also drastically improves healthcare outcomes for skin cancer: the five-year survival rate for patients with skin cancer that hasn’t spread to the nymph lobes is 99% while the survival rate is 30% for patients whose cancer has spread to distant lymph nodes (American Academy of Dermatology). Tschandl et al.’s skin cancer artificial intelligence models of clinical decision-making proved to be more accurate than relying solely on physician diagnoses and demonstrated that least experienced physicians benefit the most from AI support. However, the paper also noted that artificial intelligence has the potential to misinform and mislead seasoned professionals when left in the hands of nonexpert clinicians (Tschandl et al., 2019).
AI may seem like an easy fix for overworked or inefficient areas of the healthcare system. However, some people are skeptical about fully integrating AI into the medical world. This is due to concerns over data privacy, systems becoming vulnerable to hackers, and the difficulty of successfully implementing AI into the medical field (Khan et al., 2023). Many in the medical field including Robert Versaw, vice president of innovation and growth at Envista Holding—a dental equipment and manufacturing company—believe that though AI has the potential to revolutionize the medical field, it must be done strategically, with caution, and with enough government regulations (Versaw, 2023).
Though we are still early in unlocking and implementing the full potential of AI in the medical field, these studies show the possibilities of being able to automate many medical processes such as diagnostics. Michelle Thomson, a family medicine DO at the University of Pittsburgh Medical Center noted that an AI tool that records and summarizes patient interactions has allowed her to be “100% present for [her] patients” as a physician. Using artificial intelligence models to supplement or replace physicians’ labor in these tasks could take a load off of healthcare providers and give patients access to cheaper and more efficient care as we continue the efforts of setting up regulations and standard procedures surrounding AI in the healthcare industry.
Referenes
Sharif N, Gilani SZ, Suter D, Reid S, Szulc P, Kimelman D, Monchka BA, Jozani MJ, Hodgson JM, Sim M, Zhu K, Harvey NC, Kiel DP, Prince RL, Schousboe JT, Leslie WD, Lewis JR. Machine learning for abdominal aortic calcification assessment from bone density machine-derived lateral spine images. EBioMedicine. 2023 Jun 26:104676. doi: 10.1016/j.ebiom.2023.104676. Epub ahead of print. PMID: 37442671.
Kumar Y, Koul A, Singla R, Ijaz MF. Artificial intelligence in disease diagnosis: a systematic literature review, synthesizing framework and future research agenda. J Ambient Intell Humaniz Comput. 2023;14(7):8459-8486. doi: 10.1007/s12652-021-03612-z. Epub 2022 Jan 13. PMID: 35039756; PMCID: PMC8754556.
Janghel RR, Rathore YK (2020) Deep convolution neural network based system for early diagnosis of Alzheimer’s disease. Irbm1:1–10. https://doi.org/10.1016/j.irbm.2020.06.006
Connell GCO, Chantler PD, Barr TL (2017) Stroke-associated pattern of gene expression previously identified by machine-learning is diagnostically robust in an independent patient population. Genomics Data 14:47–52. https://doi.org/10.1016/j.gdata.2017. 08.006
Tschandl P, Nisa B, Cabo H, Kittler H, Zalaudek I (2019) Expert level diagnosis of non pigmented skin cancer by combined convolution neural networks. Jama Dermatol 155:58–65
13, A., & Robeznieks, A. (2022, April 13). Doctor shortages are here-and they’ll get worse if we don’t act fast. American Medical Association. https://www.ama-assn.org/practice-management/sustainability/doctor-shortages-are-here-and-they-ll-get-worse-if-we-don-t-act
Abdominal aortic calcification. PatientsLikeMe. (n.d.). https://www.patientslikeme.com/conditions/abdominal-aortic-calcification
Khan, B., Fatima, H., Qureshi, A., Kumar, S., Hanan, A., Hussain, J., & Abdullah, S. (2023, February 8). Drawbacks of artificial intelligence and their potential solutions in the healthcare sector. Biomedical materials & devices (New York, N.Y.). https://www.ncbi.nlm.nih.gov/pmc/articles/PMC9908503/
Lohr, S. (2023, June 26). A.I. may someday work medical miracles. for now, it helps do paperwork. The New York Times. https://www.nytimes.com/2023/06/26/technology/ai-health-care-documentation.html
Machine learning for abdominal aortic calcification assessment from ... (2023). https://www.thelancet.com/journals/ebiom/article/PIIS2352-3964(23)00241-4/fulltext
Pick, A. (2019, February 27). Aortic valve gradient - for patients with aortic stenosis. Adam’s Heart Valve Surgery Blog - Former Patient And Author, Adam Pick, Blogs About Heart Valve Replacement And Heart Valve Repair. https://www.heart-valve-surgery.com/heart-surgery-blog/2008/10/05/aortic-valve-gradient/#:~:text=Normally%2C%20the%20pressure%20gradient%20across,stenosis%20(%3E100%20mmHg.
Addressing the healthcare staffing shortage - definitive healthcare. (October, 2022). https://www.definitivehc.com/sites/default/files/resources/pdfs/Addressing-the_healthcare-staffing-shortage.pdf
Skin cancer. American Academy of Dermatology. (n.d.). https://www.aad.org/media/stats-skin-cancer#:~:text=The%20five%2Dyear%20survival%20rate,the%20lymph%20nodes%20is%2099%25.&text=The%20five%2Dyear%20survival%20rate%20for%20melanoma%20that%20spreads,nearby%20lymph%20nodes%20is%2068%25
Time is of the Essence When it Comes to Stroke Detection. Pennmedicine.org. (n.d.). https://www.pennmedicine.org/updates/blogs/neuroscience-blog/2022/february/time-is-of-the-essence-when-it-comes-to-stroke-detection#:~:text=If%20you%20are%20showing%20symptoms,you%20or%20a%20loved%20one.
Versaw, R. (2023, June 28). Council post: The transformative IMPACT OF ARTIFICIAL INTELLIGENCE IN medical tech. Forbes. https://www.forbes.com/sites/forbestechcouncil/2023/06/27/the-transformative-impact-of-artificial-intelligence-in-medical-tech/?sh=4dcec30a1ac9