Artificial Intelligence in Women's Health

Artificial Intelligence can trigger all kinds of scary thoughts, mostly the killer androids and evil computers from the movies. And the frightening pace that this technology is developing? It comes to something when the people behind AI are calling for developers to slow down.

So if you saw this episode title and were concerned about the role AI could play in women's health, I'm here to put your mind at ease!

In this episode, I describe what Artificial Intelligence is and how it learns by observing what we do. I also provide definitions of some of the key terms that you might have seen, outline some real-world benefits that apply to women's health, and much more.

Topics include:

• What Artificial Intelligence is and what it isn't
• The five ways that AI learns from humans
• Some helpful vocabulary definitions
• Where does the data come from that enables AI to solve health problems?
• How is AI currently being used in women's health?

Salsa Music extract by Liborio Conti

Don't forget to sign up for the Women's Digital Health Newsletter. 

Subscribe to the Women's Digital Health Podcast wherever you're listening right now. And please share the podcast with anyone from your community who will benefit.

Disclaimer
The information in this podcast is for informational purposes only and is not intended as a substitute for professional medical advice, diagnosis, or treatment. Always seek the advice of your physician or other qualified healthcare providers with any questions you may have regarding a medical condition or treatment.

The personal views expressed by guests on Women's Digital Health are their own. Their inclusion here does not constitute an endorsement from Dr. Brandi, Women's Digital Health, or associated organizations.

Visit Women's Digital Health and subscribe to our newsletter.

Connect with Women's Digital Health on Instagram, Facebook, LinkedIn, and YouTube.

Transcript

00:00 Dr. Brandi Welcome back to episode seven. In this episode, we'll be discussing artificial intelligence in women's health. But before we start, I want you to take a slow, deep breath. Because we're already using artificial intelligence. If you use Alexa, Siri, if you have some self-driving feature in your car, you're already using some form of AI. It's going to be okay. Welcome to the Women's Digital Health Podcast, a podcast dedicated to learning more about new digital technologies in women's health. We discuss convenient and accessible solutions that support women with common health conditions. Join us as we explore innovations like mobile health applications, sensors, telehealth, and artificial intelligence, plus more. Learn from a board certified anesthesiologist the best tips to filling some of your health experience gaps throughout life's journey. So in this particular episode, I'm going to be using a lot of dance examples. I want you to think about artificial intelligence as your dance partner with a lot of fancy technology to help it understand how to dance. For this episode, I listened to a lot of debates in preparation. I listened to experts in this space who are already aware of all of these ethical dilemmas and a lot of the challenges that we have with the data. But you know, just like when you learn to dance, and by the way, I'm always in a beginner's dance class because I have challenges just learning some of the basics. Just like dancing, you have to start with understanding the foundational concepts because I think a lot of us are just trying to wrap our heads around what exactly is artificial intelligence. So in this particular episode, we're going to define what artificial intelligence is, and we're gonna define it like we're teaching the basic steps of a dance routine. So soon we're gonna be saucing, if that's a word. We're gonna be popping, we're gonna be twerking, we're gonna be dropping it like it's hot. Whatever dance you're trying to understand, we're gonna use dance as a part of learning what artificial intelligence is. Let's take it from the top. What is artificial intelligence? Artificial intelligence is the field of teaching computers and robots to mimic. Thinking how humans think. So imagine you're trying to teach a dance partner how to move, and your dance partner is looking at you and they're trying to move like you move because they wanna look good with you on the dance floor. A dance partner that's trying to move like you move, that's artificial intelligence. It's trying to pick up all the little parts of how you dance so it can dance just like you. Now if you think about dance, there's all these types of dance, and there's all these different flavors of learning how to dance. So if you think about intelligence, there's also different types of intelligence. There's emotional intelligence, there's musical intelligence, there's physical intelligence. The intelligence that artificial intelligence is trying to be like is closely related to mathematical and logistical intelligence. So just like in dance, where you need this precise coordination and timing with the music, the artificial intelligence is trying to mimic human intelligence with mathematical and logistical components of that intelligence to be exact, the closest to how a human might think. So let's break it down a little bit more. The next question might be, well how does artificial intelligence do this? And there's about four or five different components of how it learns from humans. So let's go back to our dance analogy. Our dance partner, artificial intelligence, is watching you, and it's watching you to get new information. And the way that it watches you is through the concept of learning. So learning is the first part of how artificial intelligence is getting this information. The second part of how it gets this information is through perception. Perception is like us using our senses, like seeing, tasting, touching, smelling. We use all of those different senses to get the new information. And so artificial intelligence uses these characteristics of humans to see, touch, taste, and smell to learn this new information. The third component is reasoning. So think about a choreography in a dance routine. In order to learn this dance routine, you have to use your senses to see, to hear, maybe even touch your feet with the dance floor to get a sense of how to dance. But then the next step is, now that you have this information, what's the most logical decision to make with moving with this music? And that would be the third part, which is reasoning. Okay, the fourth component of how artificial intelligence is trying to mimic or imitate human intelligence is through problem solving. So let's go back to that dance analogy. Okay, so you're dancing, you got the moves down, and then all of a sudden the music stops. That would be a problem. And humans face problems all the time. And so you, as a human, you have to make a decision. You have to make a decision on the most effective solution to solve this problem, and that would be problem solving. Artificial intelligence imitates this component of human intelligence by making or solving a problem of what are we gonna do now that the music has stopped? Okay, this fifth component, and this is the last way in which artificial intelligence imitates how humans mathematically and logistically think is probably the most popular one, and that would be communication or language. This is the one that's hit the news lately. Language is a huge component of artificial intelligence. This is our system of talking to each other. And it's like having this conversation with your dance partner. So let's go back to our example. When you're communicating on the dance floor, maybe you're moving not only with your feet and your hips and your arms and your head, but maybe you're making facial expressions. Maybe you're creating different gestures. So artificial intelligence is imitating you by learning the different ways in which you communicate through these different sources of information. And I promise we'll get into more examples a little later on in the podcast, but understanding how artificial intelligence is trying to be like human intelligence is really important. Let's just take a break. The next part of this is introducing some vocabulary terms. And these are terms you might see in the news or you might hear in a debate, you know, and I'm just gonna introduce two or three at a time because I don't want us to get overwhelmed with terminology. But these next few terms, I think, will help you recognize what people are talking about as they go back and forth about how artificial intelligence is being used. Okay, so let's go back to our dance example. So I don't know about you, but when I was trying to learn salsa, I had to take it step by step. And we're gonna do that for all of these vocabulary terms. I'm gonna call our vocabulary term vocabvitals. And we're gonna dive a little bit deeper into the subfields of artificial intelligence just so that we understand what people are talking about. We know now that artificial intelligence is just a computer or robot mimicking the human intelligence in these different ways that we've already discussed how it does that. Another vocab term I wanna make sure we understand is neural networks. Neural networks describe algorithms or a set of instructions. And let's go back to our dance example. The goal here is for our dance partner or artificial intelligence to look as good as a human on the dance floor. So think of the algorithm like the choreography. The choreography is like a set of instructions that teaches us how to dance. So our dance partner, artificial intelligence, is using these choreographer instructions. Now, where do these instructions come from? Well, there's different sources of information. And each neural network is using one form of data. So one neural network may be using data from previous images of people dancing. Another network may be using text of previous choreographers' instructions. Another neural network may be using music sources to actually put the dance together with the music. So when you put all of these neural networks together, when they're connected, our dance partner, artificial intelligence, uses these connection of neural networks to mimic humans that use different forms of data to put the music with previous images that they have seen to make a dance. And so when these neural networks become interconnected, they're using the text, the video, the music, and they're putting it all together. Now the dance looks smooth, it looks sophisticated, it looks harmonious. And now you and your artificial intelligence dance partner, now you're ready to dance because the dance partner is using its neural networks to mimic how humans put all of their neuronal networks or the neurons in the human brain to make and learn a new dance. Okay, great, so we've covered neural networks. I'm gonna introduce one more vocab video, and that would be machine learning. So let's go to our example again. Our dance partner, artificial intelligence, has learned from these neural networks or this choreographer's instructions how to dance just like humans, but with machine learning, our dance partner has become so skilled at learning the dance movement that they no longer need the choreographer anymore. They can actually do this dance, and even when the music stops, when there's a problem, they can make a decision on their own without any guidance from a program or a choreographer's instructions to either stop dancing or keep dancing. They don't need the direction anymore, and they can perform the dance routine flawlessly. They don't need any support from a human, and that would be machine learning. So you've got three main vocabulary terms, artificial intelligence, neural networks, and machine learning. And if you understand these terms, those three terms are pretty much the foundation of having any conversation about artificial intelligence. Hey, listeners, it's Dr. Brandy. Thanks for listening to this episode of Women's Digital Health. Subscribe to Women's Digital Health on your favorite podcast platform. If you wanna know even more about how to use technology to improve your health, subscribe to our newsletter on women'sdigitalhealth.com, follow us on Instagram, Facebook, YouTube, and LinkedIn. Enjoy the rest of this episode. Okay, so I'm gonna move into datasets, and I think it's appropriate to just say, before I start talking about datasets, this is where a lot of the ethical dilemmas lie. You know, in artificial intelligence, the question always becomes, where is this data coming from? Who gets to train the neural networks or teach these neural networks how to perform any task or solve any problem? Who is teaching you how to dance? Where are you getting your information from? So whenever you hear anyone make a comment like, input equals output, this is what they're talking about. This information source that artificial intelligence is learning from is crucial. So for this particular episode, we're gonna focus on datasets in healthcare. Where is the data coming from in healthcare that powers or teaches artificial intelligence to solve problems for us? The first set of data that trains or teaches artificial intelligence would be electronic health records. So let's go back to our dancer example. You know, they need to practice their moves and they need information in order to practice these moves. They're getting this information from healthcare providers and technicians who are putting the data into the electronic medical record. So the data is being entered into the electronic medical record, and that's training the artificial intelligence to develop a solution. And typically these solutions come in some sort of prediction. So it's analyzing the data, and then it's gonna make a prediction on an outcome for a patient. They're putting all these different datasets together, like your medications, maybe your labs, maybe images, your genetic testing to help you as the patient or the doctor understand what's gonna happen next. And so using the dance analogy, having more data helps our dance partner to look more sophisticated, more flawless. And so the more information it has, the more likely it's able to predict what the next dance move is. The second source of data that artificial intelligence is learning from is medical imaging. So medical imaging would be chest x-rays, ultrasounds, CT scans. And so what's interesting about a dancer, they're studying from different images and videos in previous performances. So artificial intelligence is doing the same thing, except it's learning from all of these different images. They're taking hundreds of thousands, maybe even millions of different medical images, and these images are already marked by a doctor. Let's take a radiologist, for instance. They're looking at a medical image like a chest x-ray, and they will note that there's a mass right here. Okay, artificial intelligence looks at the mass that the radiologist has already marked and learns that whenever artificial intelligence sees on a chest x-ray from this radiologist who's already marked it, that there's a mass, over time, the more artificial intelligence images and understands what's going on, it could do it on its own. So just like a dancer who's looked at multiple videos to improve their performance, artificial intelligence is using chest x-rays, ultrasounds, and CT scans that have already been marked by a doctor or a technician to improve the way in which it identifies a mass or some problem like fluid on the lung or some foreign body without the human actually identifying it for them, although they are using previous images that have already been identified by the human. So there are many other forms of data sets that artificial intelligence is learning from in healthcare. I'm just gonna cover one more, which would be social media. So social media is used by artificial intelligence to learn how to talk about health topics. So it's teaching artificial intelligence to be conversational about people who are asking different questions about health and trying to figure out what's the best language to use to not only ask the question, but respond the best way to health topics. If we go back to our dance example, human dancers might use TikTok, Instagram, YouTube to learn about a new dance, and they are using those images or those videos to improve their dance performance. So artificial intelligence is using the same thing, same approach to learn how to communicate. Oh, great, we know our terms, we now know where the data is coming from. Now we can actually talk about how artificial intelligence is actually being used in women's health. So one of the first examples of how artificial intelligence is being used in women's health would be the ability of artificial intelligence to detect breast cancer. So we know that artificial intelligence has this ability to use medical imaging to pick up on different masses that are being identified by the radiologist or technicians over time. It can also recognize this with mammograms, which is a special form of medical imaging used to detect breast cancer. And there's some literature that supports that artificial intelligence can perform as good or even better than humans in being able to detect even smaller and smaller sizes of breast cancer that sometimes is undetectable to the human eye. Again, using the example of our dance partner, they're seeing millions of these images of dances and in this case, mammograms, and it's being used to teach artificial intelligence how to see these tumors. And it gets very sophisticated over time with seeing different images without any support from a human. Now, I just wanna emphasize here that even though artificial intelligence is using these images to learn how to detect masses, ultimately, artificial intelligence is still being used as a support tool. In other words, on its own, it's not being used to predict breast cancer. You're still gonna need the expertise of a physician to help you identify yes or no, is this actually a mass, which will help you with diagnosis and ultimately treatment. Now, another way in which artificial intelligence is being used in women's health would be in the mental health space. In this particular space, we are dancing with a very familiar dance partner, which would be Alexa, which is the artificial intelligence assistance that's used with Amazon. So Alexa is being tested as a virtual coach to help patients with mild to moderate anxiety and depression. And I found this study that is using Alexa as a coach and the platform they're using is called Lumen and they're using this chat by ability of artificial intelligence to be conversational. In other words, using the language that humans use to talk about health and really engage people who are dealing with mild to moderate anxiety and depression. And I'm keeping an eye on the literature. It looks like they've seen some mild improvements in a very small study that was majority women and that they've tested on, but they recognize they need a larger study. But Alexa is a great example of how artificial intelligence is using language or what natural language process is to become more sophisticated in how it's talking about mental health. Now, when we talk about mental health, particularly for women, we have to be very safe. We have to be very protective of how that conversation goes. So it's really important that as we see the development of artificial intelligence in mental health, that we make sure it's safe and it's supportive for all the dancers or all the patients to be able to express themselves. The last example of how artificial intelligence is being used in women's health requires us to put all the data sets that we've discussed together. And that would be in the fertility space. So in the fertility journey, artificial intelligence is helping persons who wish to become pregnant understand the best time to implant an embryo. It's also helping understand how to select the best embryo and to have a successful pregnancy. They're using a combination of electronic health records, images of previous embryos that were successful and different mobile health apps that support a person's prediction of their menstrual cycle so that they can understand when the best time to actually implant the embryo is so that you can give the person who wants to become pregnant the highest success of pregnancy. But going back to our dancer example, you can see how artificial intelligence is a skilled dancer. It's taking this data, it's analyzing it, it's making the decision on how to give the person the best chance of a successful pregnancy. And I look forward to seeing more technologies come out that use all these different forms of data to have a more precise prediction and improve the outcome for patients. Okay, I'm gonna squeeze in one more vocab vital as we come to the end of this episode. And that vocab vital would be deep learning. So deep learning is the subfield of artificial intelligence that exclusively deals with neural networks. And now that we know what a neural network is, we can compare deep learning with machine learning. Okay, let's go back to our dance example. And this time, we're going to use Beyonce dancers to compare the difference between how you might learn from machine learning and how you might learn from deep learning. Now, if you're learning how to dance with machine learning, you're using a lot of videos to learn how to dance, but it might look a little choppy. It might look a little, you might miss some details. It might take you a little bit longer to learn all of the moves. But if you were learning with deep learning, it's like having a dance coach right there. They're showing you each step and your coach not only knows your moves, but it understands your strengths, your weaknesses, and where you're missing the details. And it's helping you to improve and create even more complicated routines that's built upon previous movements that you already know. So deep learning is using all of these algorithms, all of these interconnected neural networks that learn from all of these different sources of data. It's learning not just from one video, but it's learning from different coaches and it's learning different types of dances. Keep in mind, a Beyonce dancer might not just know Beyonce dancing, but a Beyonce dancer knows tap, jazz, ballet, hip hop, salsa. They know all of these different types of dances that require a degree of detail and nuances that's a little bit more difficult to learn from machine learning. So deep learning is like having an expert coach who can look at every single detail of what you're doing and they help you to become a dance superstar, a Beyonce level of dancing. And you're using a lot more resources to build this dance performing. So machine learning is good for some tasks, but deep learning, that's next level. That's Beyonce level dancing. So that concludes episode seven on artificial intelligence in women's health. I hope that you've learned about vocabulary in artificial intelligence. You've learned about data sets and you're asking questions about how are people learning how to dance? Where's that data coming from? And you're certainly learning about how we can use the data sets to actually improve health outcomes for patients. Thank you so much for listening. I hope you enjoyed it. And if you did, please leave us a review, leave us a rating on your favorite podcast platform. Follow us on all our socials. We're on LinkedIn, YouTube, Instagram. We're finally on TikTok. I hope you enjoyed today's episode. I hope you replay it and share it with other ones so that you can enter these conversations with a lot more comfort. Keep dancing everybody. Hey listeners, thanks so much for listening to this episode. So in episode eight, we're gonna start putting it all together. We're going to listen to a real woman's journey of going through fertility challenges. We've talked to the experts. We've heard a lot about the technology that exists. And now it's important for us to put it all together. How might we better understand the problem and how might we use some of the solutions that we've been talking about in these previous episodes to help improve a real woman who's experiencing the fertility journey firsthand. Although I'm a board certified physician, I am not your physician. All content and information on this podcast is for informational and educational purposes only. It does not constitute medical advice and it does not establish a doctor patient relationship by listening to this podcast. Never disregard professional medical advice or delay in seeking it because of something you heard on this podcast. The personal views of our podcast guests on women's digital health are their own and do not replace medical professional advice.