)
Interview: How can Artificial Intelligence Improve Bias in the Health Sector?
The below interview has been created using a transcript of the video. Find out more about our 2024 AI and Healthcare agenda, or check out our pass and pricing options.
Chuck Martin
Hi, this is Chuck Martin with the AI Summit in New York City. I'm happy to say I have with me here Pablo Damascena, principal data scientist for Johnson & Johnson. Welcome.
Pablo Damasceno
Thank you for having me.
Chuck Martin
So, you just finished a program. Tell me what you talked about.
Pablo Damasceno
The talk was about using GenAI, which everyone is excited about. At Johnson & Johnson, we use it to tackle biases in our imaging systems. We focus on bringing the right drug to the right patient and ensuring that the drug works across various countries and types of patients. Additionally, we work on making sure our algorithms, built during clinical trials, don’t have inherent biases. We've been using GenAI models to remove biases throughout the clinical trial process.
Example of Bias in AI Models
Pablo Damasceno
An interesting bias I encountered in a previous job involved chest x-rays. We worked on predicting if a person had a high probability of needing a ventilator during COVID. We discovered the model was correlating the presence of pacemakers with death. The issue was that the model was identifying correlations rather than causations.
Evolution of AI in Healthcare
Chuck Martin
You've been in the AI space for about six or seven years. How have you seen it evolve?
Pablo Damasceno
AI goes through hype cycles where certain technologies become popular and then fade. GenAI is currently at the forefront. While there are many exciting applications, there are also areas that require caution. You can’t just take the results for granted—you have to test what works and what doesn’t.
Role of AI in Healthcare's Future
Chuck Martin
What do you see as AI’s role in healthcare going forward, given the privacy challenges?
Pablo Damasceno
We believe it’s the future. One major issue in healthcare is that doctors are overloaded, and their biases can influence diagnoses and prescriptions. At Johnson & Johnson, we work a lot with rare diseases where misdiagnoses are common. Regulatory agencies like the FDA are interested in AI because they know AI systems may reduce these biases and improve accuracy.
Doctors' Reception of AI
Chuck Martin
Are doctors open to AI?
Pablo Damasceno
Some are initially hesitant, wondering if they'll lose their jobs. However, once they understand how the models work, they realize AI can assist them in making more informed decisions. AI highlights potential diagnoses, like pulmonary hypertension, allowing doctors to make more accurate choices. It's about humans working with AI, not being replaced by it.
Adapting to AI in Healthcare
Chuck Martin
Is it a learning process for doctors?
Pablo Damasceno
Yes, it’s about adapting and becoming more familiar with the technology. Eventually, it will become like using Google—you won't think about whether AI is involved or not. For example, chest x-rays will come pre-annotated with AI analysis, and doctors can decide how to use that information.
AI Integration in Healthcare and Clinical Trials
Chuck Martin
Is that how you see AI, integrated and behind the scenes?
Pablo Damasceno
Yes, that’s the goal, especially in healthcare. We’re working towards helping doctors make better decisions. In clinical trials, AI can make the process faster, cheaper, and more efficient. AI can also help ensure drugs work for the right patients by identifying which populations respond best to certain treatments, avoiding wasteful clinical trials.
Future Conversations on AI Regulation
Chuck Martin
A year from now, if we’re having this conversation, what will we be talking about?
Pablo Damasceno
I hope we’ll be discussing regulatory progress. The conversation around AI regulation is ongoing and slow, but a year from now, I think we’ll see more applications. For instance, we’ll have expanded the use of GenAI for imaging, such as chest x-rays and endoscopy, into more areas.
Find Out More
Stay at the forefront of AI healthcare trends, innovations and solutions by registering for our 2024 show.
Or, see more highlights from the 2023 New York AI Summit.
){kind=logo}
){kind=logo}
){kind=logo}
){kind=logo}
)
)
){kind=logo}
){kind=logo}
){kind=logo}
)
)
){kind=logo}
){kind=logo}
){kind=logo}
){kind=logo}
){kind=logo}
){kind=logo}
){kind=logo}
){kind=logo}
){kind=logo}
)
){kind=logo}
)
){kind=logo}
)
){kind=logo}
)
)
-(1).png/fit-in/500x500/filters:no_upscale()){kind=logo}
){kind=logo}
){kind=logo}
)
.svg){kind=logo}
)
)
){kind=logo}
.jpg/fit-in/500x500/filters:no_upscale())
){kind=logo}
.jpg/fit-in/500x500/filters:no_upscale())
)
)
){kind=logo}
)
){kind=logo}
){kind=logo}
)
)
)
){kind=logo}
){kind=logo}
){kind=logo}
){kind=logo}
){kind=logo}
.png/fit-in/500x500/filters:no_upscale())
)
.png/fit-in/500x500/filters:no_upscale())
){kind=logo}
){kind=logo}
.png/fit-in/500x500/filters:no_upscale())
