danielle-n-knapp-4 - Tumblr blog

danielle-n-knapp-4 · 1 year ago

Text

Artificial Intelligence in the Medical Field

It is now the year 2024, if you need to see a doctor you can do it over the phone or via video chat, or even talk to a computer and have that diagnosis you. Alexander Ávila a 23 year old created a video discussing how he thought he had autism, he went online and found a credible source that gave him a test to complete and sent him his results. He in fact did have autism. Which is wonderful for him but what happens when AI (Artificial Intelligence) gets deeper into the medical field? During the 2020 pandemic, which we all loved so much, whenever you thought you might have covid or just felt sick in general, you would go get tested. For some people including myself, would use Dr. Google or another app to help narrow down what you might have. During Covid an app called K health was created. This app allows you to put in your symptoms and a computer (AI) which will ask you questions and based on the symptoms you give them they can narrow down to what you might have. The founder of the app created it so that “if you wake up in the middle of the night with a stomach ache or are on vacation and your kid gets a fever, you shouldn’t have to wait to get care.” Which is reasonable except for when someone like myself, who has a nervous system disorder, clicks in their symptoms, and gets told that they at the age of 20 are having a heart attack when in reality it’s just the disorder they have acting up and me not knowing it.

Now getting diagnosed with something on an app isn’t always that bad but they can sometimes get it wrong. And when it comes to telling the difference between the flu, a cold and covid is not something most people will be extremely concerned about. But what if you get

admitted to the hospital and need to get a scan done (MRI, CAT, CT, X-ray…ect.,) and AI is the one reading it and not a living person? For those who don’t know but when you get a scan done a person with a medical degree in radiology reads your scan. AI however is being trained to read these scans. The reasoning for this is AI can review the scans quicker than the radiology and doctors. I understand it can be frustrating to have to sit and wait for a doctor to come tell you your results. However, computers are not trained for everything, meaning that just like humans they can get things wrong. So even if AI reads your scans someone (hopefully) will still have to read them and make sure the computer was correct, so is it faster?

Now AI is not a horrible thing, AI can help physicians and nurses and medical staff by altering them to high or low lab values when documenting, telling you if you are how many people in a city, state, country have a disease by tracing and AI making a map a lot faster than humans can. I’m just discussing the drawbacks of implementing AI. With that said AI will not cost anyone their jobs because ultimately, we still must double check it’s work but we could get some stuff done faster. The main drawbacks are data collection, algorithm development, Ethical concerns, social corners, clinical implantation concerns, biased and discriminatory algorithms. Let's start with data collection, and DeepMind is an AI for healthcare rolled out by google in 2018, they needed 1.6 million patients to just create that. Which means that to have AI help in the medical field they would need at least that many people to be willing to allow their medical data to be used. And I say allow because everything in the medical field requires consent. The next big one I would like to discuss is biased and discriminatory algorithms, in the healthcare system with AI. People are trained to be as unbiased as possible while in nursing school, medical school, so on and so forth. AI would also have to be trained to not be biased but would also have to be trained to know that people of one sex are more likely to get cardiovascular disease, and some races are more likely to develop heart disease, but it would have to be an unbiased training. Sounds a lot like a double negative, you can't be bias, but you need to know that a person of African American descent is more likely to develop heart disease, while a person of European decent is more likely to develop problem with the circulatory system but can’t assume that is what the person has before you know all the information.

There is a future for AI in the medical system. There are benefits to AI, although I personally do not want it more involved in healthcare due to potential compilations. They talk about how AI will be an assistant to those working in the hospital or doctor’s office. At a local hospital they are using a system that automatically adjusts your iv pumps to run according to the amount of medicine in the bag. But they have been having problems due to integrating them, it takes time and if the nurse is more worried about if the pump hooked or synced to the computer then they could be missing out on information with their patient and miss learning new critical data. Which I think raises questions about the ethical practice and safety of the patients. Now before I end, I want to just add a myth that has been busted about AI in the healthcare system. AI will replace clinicians, in fact clinicians will not be replaced and can hopefully use AI to help advance their ability to care. Meaning, AI will be integrated in a way that will allow for your doctor to better assess what is happening to you, an explain of this is when a nurse documents something and a range is low or high it will notify the nurse and doctor instantly to allow them to determine why it is high or low and the best route for care. There is another myth about programming that says you don’t need to officially program the computer but need to teach the computer which is what google has done. So, is AI really that helpful and should it really be allowed in the healthcare system?

0 notes