AI and Healthcare by Rebeca Salmon

Personalized Overview

rebecasalmon00 — 2025-01-18 01:22:52 UTC

Physician's First AI Consultant

rebecasalmon00 — 2025-01-18 01:55:54 UTC

The 1972 brainchild of Stanford University, MYCIN is a physician's consulting system that prescribes antibiotics for bacterial infections. Its programming was founded on a bank of physician knowledge, or "rules" by which it were to operate. The device would ask the physician a series of questions, possibly patient information, to arrive at a diagnoses. The device when tested again hundreds of practicing physicians was found be effective. Achieving results that were virtually no different from actual physicians.

This is a machine learning technology undoubtedly has the potential to save lives, but researchers at Columbia University have uncovered some limitations that are concerning. The information programmed into MYCIN is taken as fact, with is a rigid framework for the ever-changing nature of medicine. It has the tendency to over-prescribe medication as it often considers several potential bacteria that could be detected as opposed to a conclusive diagnosis. It doesn't consider the caliber of doctors inputting information, as well. Which may have impact in the quality of the "facts" being adopted into the system. For this reason, MYCIN would fall out of popularity and is not longer used today.

AI and MRI

rebecasalmon00 — 2025-01-18 02:13:19 UTC

According to the NYU Langone Health Research Institute, AI has made strides in MRI research. MRI scans, which are traditionally lengthy, are being outperformed by AI imagery, which can yield the same results in less time, with better quality and less information. Experiments were performed comparing a standard MRI to a Rapid AI scan that obscured parts of the image. The Rapid AI was able to accurately generate parts of the missing imagery, similar to MYCIN, it utilized information from previous lab results to construct the image. As a development, more time is needed to accurately assess the full potential of AI and MRI, and although promising, like MYCIN, is still only used as a consulting tool, not a replacement for the specialists as a whole.

AI and the Road to Vaccination

rebecasalmon00 — 2025-01-18 02:34:13 UTC

A recent development of AI within health has directly impacted millions across the globe. The creation of a vaccine is a long, and arduous process to ensure the safety of future patients. Before release to the general public are extensive tests, or trials that record the drug performance on the body, side effects, etc. The results are analyzed by teams of doctors, who after extensive deliberation, determine whether or not the drug is suitable for consumption. Although heavily oversimplified these details alone are evidence of how difficult the process is. If that is the case, then how were physicians able to pump out the COVID-19 vaccine so quickly? Of course, the urgency, and severity of the virus would push for more funding for the research, but AI had its hand in it as well. AI technology was used to compile the results from the clinical COVID trials, significantly cutting down on time, and thus enabling vaccines to be manufactured so quickly. However, according to an article by the Harvard Business Review, AI was unable to reach its full potential as they were of course outperformed by real physicians in a real setting. They do not consider unique patient details but rather work by sorting similarities, but demographic differences for example may be more useful to the story than the AI can detect. A recurring theme with these AI technologies is that they do not effectively replace physicians but can serve as an aide to their work. That is no different in this realm, and why it does not take on a more aggressive role in healthcare.

Future Trends and Ethical Considerations

rebecasalmon00 — 2025-01-18 02:52:06 UTC

As discussed earlier, although AI is making its way into healthcare, it is a far cry from being the sole, or even leading authority within these spaces. Their primary role within healthcare for the last fifty years has been that of an assistant to physicians in their respective fields. The learning technology is based on the input (research, lab results, etc.) of the physicians themselves which may lead to concerns of accuracy: is the information being input by someone qualified, how can we vet this information (what number of physicians is sufficient/qualified for accuracy), how often must information be updated? Additionally, ethical concerns arise in the storage of such data, how can patients ensure that their appointments, surgeries, and other medical data are being utilized consensually? Is it possible to revoke consent at a later date, once integrated into the systems? Issues of malpractice as a result of over-reliance on these systems can lead to issues in the field as well. All of these issues lead me to believe that the current role of AI within healthcare will remain relatively unchanged in the years to come.

Societal Impacts

rebecasalmon00 — 2025-01-18 03:01:30 UTC

Aside from the practical issues of AI leading in the workplace, there are of course the social implications of a healthcare system that chooses to rely extensively on AI. As mentioned before issues concerning consent as it pertains to AI pose a threat to patients seeking medical care. It is unclear whether AI can discard input information if patients choose to revoke medical information provided in the future. And if so, to what extent will it affect the data collected, which is based on the acquired information? Additionally, when discussing the input of information, vetting information coincides with bias. Who is authorized to input information and how many physicians are necessary to achieve accurate results and a non-biased information base? Within the healthcare field persons of color are at the highest risk of receiving poor treatment on the grounds of bias. If we cannot ensure fairgrounds in healthcare without AI, its addition would further perpetuate bias. Although not an immediate threat to job security in the healthcare field, as it is generally a consultative tool, it can cut costs, because of its speed. Finally, form an environmental perspective, large-scale AI, ie collecting information via data centers, contributes to the excess emission of hazardous waste, and the over use of water, and electricity, further endangering the planet.

Rereferences

rebecasalmon00 — 2025-01-18 03:12:03 UTC

AI has an environmental problem. here’s what the world can do about that. UNEP. (n.d.). https://www.unep.org/news-and-stories/story/ai-has-environmental-problem-heres-what-world-can-do-about

Outlines effects that AI has on their environment and the dangers of normalized use.

Chakravorti, B. (2022, March 17). Why ai failed to live up to its potential during the pandemic. Harvard Business Review. https://hbr.org/2022/03/why-ai-failed-to-live-up-to-its-potential-during-the-pandemic

Discusses the shortcomings of AI in healthcare against real physicians, as it is related to COVID-19 (the need for speed is not worth sacrificing accuracy)

Ethical and social implications of AI use. Ethical and Social Implications of AI Use | The Princeton Review. (n.d.). https://www.princetonreview.com/ai-education/ethical-and-social-implications-of-ai-use

Discrimination in healthcare, bias

Medium. (2024, July 31). Ai in healthcare: Innovations and future trends. Medium. https://medium.com/@aitechdaily/ai-in-healthcare-innovations-and-future-trends-f556899bd288

General info on new AI innovations in medicine

Recht, M. P., Johnson, P. M., & Sodickson, D. K. (n.d.). Artificial intelligence reconstructs missing data from Rapid MRI scans. NYU Langone News. https://nyulangone.org/news/artificial-intelligence-reconstructs-missing-data-rapid-mri-scans

AI and MRI Background Information

Shortliffe, E. H. (n.d.). Rule-based expert systems: The MYCIN Experiments of the Stanford Heuristic Program. https://people.dbmi.columbia.edu/~ehs7001/Buchanan-Shortliffe-1984/Chapter-05.pdf

MYCIN Background information (OG source)

Wikimedia Foundation. (2024, November 18). Mycin. Wikipedia. https://en.wikipedia.org/wiki/Mycin

MYCIN Background information

Reflection

rebecasalmon00 — 2025-01-18 03:30:29 UTC

Overall, this research project has piqued my interest in AI in the healthcare field enough to stay informed on the topic long term. I was able to gain a new perspective on AI; the extent of its presence in the field was overestimated. While able to appreciate its speed, and cost-effectiveness concerns surrounding accuracy, bias, and other societal concerns, continue to make me wary. It is evident however that AI will not make any major leaps to the forefront of healthcare, it is interesting to see how physicians will utilize the tool to their advantage. As long as physicians remain adamant about keeping humanity at the forefront of healthcare AI can continue to be a useful helper in the field.

Overview Video

rebecasalmon00 — 2025-01-18 03:56:14 UTC