They also highlighted insufficient staff, cost of accessing treatment, lack of investment in preventative health, and aging population as some other problems they face.
There’s an abundance of problems to solve in the world’s healthcare systems. And guess who loves solving them? Dr. AI — promising to solve problems faster than humans they ever could — and its friends, including Google and NVIDIA.
Without doubt, AI has become synonymous with automation, seamlessly performing time-heavy tasks in record time and with little to no human intervention. This has proved to be a useful function for many healthcare businesses.
“As we specialize in pulmonology, the pandemic was pretty devastating for us. Everyone was burnt out, and even the simplest tasks took a lot of time. These simple tasks are mostly administrative processes, such as documentation,” Christine Kingsley, a US Advanced Practice Registered Nurse (APRN) and Lung Institute’s health and wellness director, told HubSpot.
Leveraging AI tools for automation was a game changer for Lung Institute, especially after the COVID-19 pandemic.
Acknowledging her industry’s preference for traditional, tried-and-true techniques, she adds, ”Don’t be afraid to try out AI tools at your disposal because it’s a good first step to reducing the gap in the healthcare industry.”
Another core use case of AI in healthcare is in medical imaging.
Conventional analysis of medical images like MRIs, CT scans, ultrasounds, radiography, X-rays, and more could take a human professional anywhere from a few minutes to several days depending on its complexity.
But with AI-powered algorithms, you can analyze images with impressive accuracy and speed, leading to improved patient outcomes.
This is why many healthcare businesses are today focusing on AI-assisted medical imaging, a market projected to reach over $840 million by 2030.
Next, there’s personalized medicine.
No two human beings are genetically identical, not even monozygotic twins. A combination of this genetic makeup, lifestyle choices, underlying healthcare needs, etc. mean different medical treatments for different people.
Leveraging the massive data AI is built on, this technology can analyze a patient’s medical history, family medical history, genetic information, and other essential factors to predict disease risks and recommend treatments tailored to the patient’s specific needs and preferences.
In a Forbes article, Renen Hallak, founder and CEO at VAST Data, believes AI will lead to the “democratization of personalized medicine.”
“Personalized medicine shifts the practice from statistics, the law of averages and hopes and prayers for the best outcome to more precisely and explicitly identifying abnormalities within the individual patient,” Hallak writes. “Over the next several years, I believe we’ll reach a point where an individual’s health will be nearly pristine.”
Then there’s also telemedicine, which is the use of telecommunication technology to provide healthcare services in situations where distance is a problem.
Its application can range from asking a medical service provider via text how many grams of tylenol a pregnant person can take in one day to performing an appendectomy through remote surgery, otherwise known as telesurgery. Yeah, that’s real.
With AI integrated into this, diagnostic accuracy, remote patient monitoring, teleconsultations, and personalized care can improve patient outcomes.
Generative AI also comes incredibly handy in the drug discovery and development departments, from target identification and validation till the drug hits the market.
Where traditional drug discovery is reliant on trial and error experiments that are uber-expensive and may take donkey’s years to enter the market, GenAI-powered drug discovery is cost-effective and boasts faster processing times.
On average, it takes ten years and $1.4 billion to bring one single drug from lab to market.
According to McKinsey. AI-assisted drug development will result in:
Now let’s talk about who’s driving this innovation.
There are many AI-powered companies leading the revolution of healthcare in this new age, applying AI to solve some of healthcare’s most difficult problems.
Arterys uses imaging technology to simplify diagnosis of heart defects in newborns and children, while Google Health is developing AI models like the Med-PaLM 2 for healthcare.
Another AI-powered company Butterfly Network makes ultrasound scans more accessible to underserved communities with its hand-held whole-body imager and AI image interpreter.
The work is quite impressive.
On the backend, however, are AI hardware manufacturers like NVIDIA.
Here’s some context.
Away from the central processing units (CPUs) that have powered computing for as long as we can remember, graphic processing units (GPUs) create a new playing field where functions like complex mathematical computations and parallel processing come to dance.
These attributes are non-negotiables for advanced technologies like AI, making GPUs ‘AI Boom’s Most Indispensable Prize’ according to the New York Times.
Controlling over 80% of the current GPU market, NVIDIA is what you’d call top of the GPU line. Still, at the GTC Developers Conference in March, NVIDIA’s co-founder and CEO Jensen Huang unveiled the company’s latest AI chip in a two-hour keynote address.
“A very big GPU,” Huang called it in his address.
The design of NVIDIA’s GB200 Grace Blackwell Superchip “enables organizations everywhere to build and run real-time generative AI on trillion-parameter large language models at up to 25x less cost and energy consumption than its predecessor,” the company said in a press release.
What this means is that everything AI can currently do in healthcare, it’ll do at even greater speed, lesser cost, and flawless accuracy.
“Blackwell’s breakthrough technological capabilities will provide the critical compute needed to help the world’s brightest minds chart new scientific discoveries,” Demis Hassabis, co-founder and CEO of Google Deepmind, added in the press release.
The conference also saw two other announcements with implications for AI in healthcare.
First on the list are the generative AI microservices launched by NVIDIA Healthcare to advance drug discovery, medtech and digital health.
These microservices, 25 of which were launched on that day, would allow healthcare providers access some of the latest capabilities of GenAI technology from anywhere in the world and on any cloud.
These small, independent services made available to researchers, developers and practitioners in healthcare have capabilities that range from near-infinite drug compound screening, smarter digital assistants, and optimized patient data for better disease detection.
Kimberly Powell, vice president of healthcare at NVIDIA, says this is the first time in history that “we can represent the world of biology and chemistry in a computer, making computer-aided drug discovery possible.”
Then there’s Project GR00T, a foundational model for humanoid robots — literally every American sci-fi blockbuster movie come true.
Instead of taking over the world though, the robots created with this foundational model will be able to “understand natural language and emulate movements by observing human actions — quickly learning coordination, dexterity and other skills in order to navigate, adapt and interact with the real world.”
In the delivery of healthcare services, a humanoid, without the physical or mental limitations of a human being, could really do wonders. Who knows, robots may one day perform invasive surgeries without human interference.
Still, it’s not all rosy on this side of Nirvana.
The major concerns about AI in healthcare are two-pronged.
On one hand is confidentiality and patient data privacy. There are not many things an average person considers as confidential as their medical history. Nobody wants the next person to know about that nasty rash they had to treat last summer.
It’s why doctor-patient confidentiality exists. But for these AI models to function as they should, they need vast amounts of data, which would typically include diagnostic images, genetic information, medical records, and more.
This, unfortunately, makes it a lucrative target for hackers. Through a breach notification portal, the Department of Health and Human Services (HHS) tracks healthcare data breaches affecting at least 500 individuals.
It’s only April of 2024 and nearly 250 breaches of healthcare data have already been posted on the portal.
The main cause? Hacking, which has so far remained a primary cyber-threat in healthcare.
With AI and its large packets of data, confidential patient information can quickly become prey lying in wait.
The second concern is reliability.
60% of American adults say in a Pew Research report that they’d feel uncomfortable if their healthcare provider relied on AI to deliver said services, especially in diagnosing diseases and recommending treatments. And this discomfort is well-founded, especially when you consider concerns about the transparency and accountability of AI algorithms, as well as fears of biases.
Dr. Bertalan Mesko, director of The Medical Futurist Institute and one of LinkedIn’s top voices in Healthcare, conducted a poll on his personal LinkedIn page asking his audience if they believed it was ethical to use AI in making life-or-death healthcare decisions without human oversight. Of the 833 responses the poll got, over 700 said a big, fat NO.
Is it all doom and gloom for AI in healthcare then? Absolutely not.
While the reception of AI is likely to improve with continuous improvements and familiarity with AI-driven healthcare solutions, concerns about its safety is a different ball game.
However, like all AI applications across different sectors, the existence of a tried-and-tested ethical and regulatory framework for the adoption of this technology in healthcare as well as compliance to the same could mitigate these concerns.
For example, the World Health Organization (WHO) this year released a paper titled Ethics and Governance for Artificial Intelligence in Health: Guidance on Large Multi-modal Models (LMMs) to help member states of the United Nations understand the pros and cons of leveraging LLMS in health care.
This is in addition to regulatory obligations like the Privacy, Security and Breach Notification Rules of HIPAA, which aim to protect individuals’ health information and ensure its confidentiality, integrity, and availability within the healthcare system.
The regulation exists, but making certain the required entities comply with them is an entirely different issue.
So, as far as the question goes, AI in healthcare does hold great promise. But it would take even greater work to fully realize its potential.
