AI in Healthcare and How to Define Its Foundation for Ethical Concepts.
-Maria Wellman
#Artificial intelligence (#AI) is starting to progress into the healthcare industry; we can't stop it, and why should we? It will save us all time, money, and possibly our lives.
Right now, the potential of AI is only hindered by our imagination. AI could help doctors reach diagnoses in difficult medical cases, perform flawless surgeries, or scan medical journals for treatments. Whatever AI is applied to, it could solve critical issues that have held back the progression of healthcare innovation.
Where is the crack in the current #healthcare system?
There are many flaws within the current US healthcare system and systems worldwide. Many of these issues could be mended with AI.
One of the biggest upcoming burdens on the system will involve the increased demand for healthcare providers. Projections indicate that by 2030, the last baby boomers will be hitting the low end of the retirement age, which means about seventy million more people will simultaneously need care. Some of those retiring are currently in the medical field, therefore, the supply of healthcare workers will decrease at the worst possible time.
Let's discuss how to solve this issue using what we have now. The goal is to increase the supply of healthcare workers. Unfortunately, the issue's existence isn’t because there is a lack of desire to go into the healthcare industry. The influx of people trying to study is being hindered on another level.
According to AACN’s report on 2023-2024 Enrollment and Graduation in Baccalaureate and Graduate Programs in Nursing, U.S. (Rosseter, 2023), these schools turned away almost 67 thousand qualified applications from baccalaureate and graduate nursing programs in 2023. Now, why did they do this? There are not enough funds to keep schools running properly with any more students than they have currently, which also leads to them not having enough faculty members or space.
In order to fix the healthcare provider population disparity, the educational system would need to be fixed as well. Two of the biggest challenges in the country, very daunting challenges considering how little time we have to solve this.
How can #AI help our #healthcare system?
AI in healthcare could benefit everyone in so many ways, as long as we take the leap of faith together. There are risks, of course, and people are trepidatious about the idea of AI helping us in such an invasive way, but if we do it carefully and ethically, then we could be on the cusp of something great.
AI could lead to improved diagnoses and assist providers so they have more time to care and less administrative stress, therefore helping their accuracy and work. AI can analyze large sets of medical data in a way that is faster and more thorough than a human. This alone would be a great stepping stone when creating improved ailment detection methods. AI also allows better personalized treatment and accuracy, it can be helpful in early disease detection and projecting potential risks in a patient's future health.
When I was younger, I went to a museum and saw a video presentation on how doctors could perform surgery with different types of technology that allowed even the most microscopic of procedures to be completed safely. This can be taken even further; AI could assist surgeons with procedures that would allow reduced errors and faster and more efficient procedures.
These integrations sound simple enough on paper, but how do we get there? More importantly, how do we reach these goals safely?
Safety is paramount when discussing AI’s integration into healthcare, and there are different thoughts about how that can be implemented. One word that swirls around when talking about AI safety is “ethics.”
What is “#ethical” when applying the term to #AI?
Some people think ethics refers to data security, patient privacy, and unbiased assessment of data regarding race and gender; others think ethics refers to how the actual mind of the AI model functions. How it treats patients and whether or not it has the potential to hallucinate
In my opinion, there are three categories that AI should be judged by to determine its total ethical safety. One is the human implementation of AI, two is what the AI is trained on, and three is how the AI behaves.
I. Human #AI Implementation
How the AI is implemented determines the intention and is the basis for its construction. The model and system should be registered as a medical device and have data storage approved by HIPAA. Another important aspect would involve regulation and monitoring, how the model handles patient information, and data security. This mirrors the regulations that hold providers accountable, so it should be the same for the AI.
II. What the #AI Model is Trained On
What the AI Model is trained on is the most important part of the equation because it has to be trained on accurate and current medical data, but it also has to be trained on how to be unbiased, how to reveal information, and how to retrieve data accurately. Transparency is key here and constant monitoring is necessary. The use of an LLM judge would not be helpful, I find it incredibly redundant to have an AI judge another AI; only humans can handle the training to prevent hallucinations.
III. How the #AI Behaves
A doctor follows an ethics guide, like the Hippocratic oath, and is bound by well-crafted wisdom and specific job expectations. They are harshly judged on how they do their job. All of these facets can affect whether or not a doctor can practice or keep their job, and the same pressure should be put on the AI medical system.
It only makes sense to have AI in the mechanism we have already created to monitor and keep doctors ethical.
Determining how an AI model behaves is done through training. The model is fed specific data curated to create specific results; it can be helped with supervision, and as the model progresses, it can be adjusted.
The human brain has cognitive functions, and we should craft and define AI that way. We all need critical thinking, abstraction, understanding, and a way to apply knowledge and make logical judgments, and so does an AI model.
A great way to define and label concepts concerning AI is through definitions we know well enough to quantify.
From there, we can break down AI learning.
AI IQ- the ability for the AI to learn
AI EQ- the ability of AI to understand the emotional aspects of a solution
AI Wisdom- what the AI learns over time and then applies to problems
If these three processes are integrated into AI as a foundation and can be programmed so they can function within the accuracy realm of a basic human, then we will have unlimited potential and a brighter future.
