AI in Healthcare – Part 2

As I mentioned in “AI in Healthcare – Part 1”, the use of AI in healthcare has roots going back into the 1950s alongside the development of the field of AI itself. Progress was limited for decades, but now things have changed dramatically. A handful of factors ignited the current boom. The introduction of awesome computer power from Nvidia was the fuel. A reliable and fast Internet emerged. The massive datasets and complex calculations required for AI training and operation got the needed boost.

The allure of AI has sparked an influx of talented students. Top institutions are offering cutting-edge theoretical knowledge and are now converting it to practical applications through collaborations with leading AI companies. The perfect storm of advancements is now enabling AI to process information at unprecedented speeds and uncover hidden patterns within the data, leading to breakthroughs in many fields. Healthcare is being viewed as a major player. Figures from 2022 and 2023 indicate a significant portion of VC funding is directed toward healthcare. Some sources say, healthcare-related ventures might attract somewhere between 60% and 70% of VC investments.

I am most interested in the AI impact in healthcare. Healthcare is so far behind. Dr. Paul Wright, Senior Vice President and System Chair for the Neuroscience Institute at Nuvance Health, reminded me of a great example: Delta Airlines scans your face and says welcome aboard. Checking in for a healthcare appointment is being greeted with a clipboard and asked to take a seat. AI can help healthcare catch up with other industries. The good news is there is a lot going on with AI in healthcare. So much so, The New England Journal of Medicine now offers a weekly email newsletter called “NEJM AI This Week.” The newsletter highlights continuously published articles, features, podcasts, and events.

I believe AI will accelerate the transformation of various aspects of patient care and medical research. This will lead to more affordable and accessible healthcare for all Americans. In this blog post, I will highlight AI progress in two areas.

AI is transforming the field of radiology, making it faster, more accurate, and beneficial for both radiologists and patients. First, AI is enhancing image analysis and detection. AI algorithms are trained on vast amounts of medical data, including images like X-rays, MRIs, and CT scans. This allows an AI to assist healthcare professionals in identifying patterns and anomalies that might be missed by the human eye, potentially leading to earlier and more accurate diagnoses of diseases like cancer. It automates repetitive tasks like identifying structures or lesions in scans, freeing up radiologists for complex cases. AI can also provide second opinions and insights to support doctors in their decision-making.

A study published in Diagnostic Imaging in 2023 focused on the ability to detect abnormalities in chest X-rays. Researchers compared an AI system (ChestLink version 2.6) to radiology reports from human radiologists. The AI achieved a significantly higher sensitivity rate, meaning it was better at correctly identifying abnormal X-rays (99.1% for AI vs. 72.3% for radiology reports). Overall, AI in radiology has the potential to lead to faster diagnoses, improved accuracy, reduced healthcare costs, and ultimately, better patient outcomes. However, it is crucial to remember AI serves as a tool to empower radiologists, not replace them.  Ultimately, qualified medical professionals will continue to make final diagnoses and treatment decisions.

An area with huge potential in healthcare is in drug discovery and development. Google is putting a lot of money into this area at its Google DeepMind. DeepMind began as a London-based AI research lab in 2010, founded by Demis Hassabis, Shane Legg, and Mustafa Suleyman. In 2014, Google acquired DeepMind, merging it with their own AI efforts under the name Google DeepMind. Their mission has been to develop powerful artificial intelligence to solve complex problems and expand scientific knowledge. DeepMind has made significant contributions in various areas, including protein folding, games, and scientific discovery.

DeepMind’s AlphaFold program has become a leader in predicting protein structures, a crucial step in understanding protein function leading to the development of new drugs. DeepMind’s AI programs have mastered complex games like Go and StarCraft II, pushing the boundaries of AI decision-making and strategic thinking. In scientific discovery, DeepMind developed AlphaTensor, which can discover new and efficient algorithms for complex mathematical problems.

The area set to have major impacts in healthcare is Google DeepMind’s AlphaFold AI model which has been a game-changer in understanding proteins, the building blocks of life. This week, they released a new version, AlphaFold 3, with a significant leap forward. AlphaFold 3 can now predict how almost all biomolecules interact with each other. This is profound. The biomolecules include proteins, DNA, RNA, and even small drug molecules.

The ability to predict interactions has the potential to revolutionize various fields, especially in healthcare. New drugs could be designed more efficiently by understanding how they interact with their targets at a molecular level. AlphaFold 3 represents a significant advancement in our understanding of life’s fundamental building blocks and their interactions, with the potential to unlock breakthroughs in medicine. I expect positive surprises over the next few years.

Next week, I will discuss progress with AI in personalized medicine, administrative tasks and chatbots, robot-assisted surgery, improved accuracy and early diagnosis, and overall enhanced efficiency and productivity. It’s important to note AI is a tool which complements healthcare professionals, not replaces them.  While AI offers substantial benefits, ethical considerations and potential biases require careful attention in its development and implementation.

Note: I use Gemini AI and other AI chatbots as my research assistants. AI can boost productivity for anyone who creates content. Sometimes I get incorrect data from AI, and when something looks suspicious, I dig deeper. Sometimes the data varies by sources where AI finds it. I take responsibility for my posts and if anyone spots an error, I will appreciate knowing it, and will correct it.