This Content Is Only For Subscribers
One of the most exciting aspects of AI is its implications for healthcare. Today, doctors and other medical professionals routinely augment their human skills and experience with the help of intelligent machines.
These machines can process information (including images) and generate data-driven predictions with incredible speed. But they still lack human qualities like empathy and compassion that will always be essential elements of providing healthcare.
Nevertheless, the application of AI is transforming day-to-day interactions between patients and professionals, as well as leading to new breakthroughs in medical science and drug development.
Of course, the use of AI in healthcare raises more ethical questions than it does in just about any other field. If its use isn’t carefully considered, constantly assessed, and, when necessary, legislated, then bad data and bad algorithms both have the potential to negatively impact patient outcomes.
With that in mind, let’s take a look at some of the most exciting and potentially life-saving ways that AI is being used today, from routine patient appointments to the frontier of medical research.
How Is AI Used In Healthcare?
Starting with the day-to-day, AI can augment the decision-making of doctors, surgeons and nurses as they assess patients and consider treatment options. Clinical decision-support systems don’t replace the need for the doctor to make a decision. Instead, they ensure the doctor has all the information they need and is aware of all of the options available. They can suggest medication and remind the doctor to consider particular factors, but the human healthcare provider will have the final say.
Professionals are also increasingly making use of AI-powered virtual assistants and chatbots to help with tasks like scheduling appointments, creating prescriptions and filing notes.
A large number of use cases center around computer vision – using AI to understand images captured by cameras. Here, it’s used to interpret medical images more quickly and accurately, for example, detecting indicators of cancer in X-rays or MRI scans. For example, studies have shown that AI enables faster and more accurate detection of early indications of multiple sclerosis by examining MRI images.
Surgeons can use AI-driven AR systems to assist as they operate. Tools like the HOLO Portal augment their ability to understand what they see when they look inside us, offering real-time information and assistance.
And for patients outside of the hospital, assistants and wearable devices can provide information, reminders to take medication, remote patient monitoring, recording vital signs, and raising an alert if human intervention is needed.
In the laboratory, AI is playing a huge role in discovering and seeking approval for the use of new drugs. In fact, the number of INDs – the process used when investigating a new drug – launched in the US that involve AI discovery has increased tenfold in the past two years.
Successes include Insilico Medicine, which recently entered phase 2 clinical trials for a new drug to treat pulmonary fibrosis, discovered with its Pharma.AI tool. And Etcembly has used generative AI to create an immunotherapy drug for cancer treatment.
One of the most exciting applications of AI in healthcare is the opportunities it is creating in the field of personalized medicine.
By analyzing the genetic profiles of individuals, AI models can build an understanding of us in order to predict vulnerabilities to conditions and illnesses as well as our response to different treatments.
Beyond this, AI can also build models and simulations that help us to take a holistic view of a patient, considering diet, exercise, and environment, as well as the effects of specific illnesses.
One great breakthrough that’s already been achieved with AI is the development of AlphaFold 2. Developed by Google Deep Mind AI lab, its purpose is to predict protein structures formed when cells divide and form new cells in the human body. This creates the possibility of tailoring medicines to a person’s unique biological makeup in order to fight disease while minimizing harm to the patient. Many use cases have been found for protein prediction models like Alphafold, from research into new antibiotics to combat antibiotic resistance to accelerating new treatments for rare tropical diseases.
The advent of generative AI – heralded by the arrival of tools like ChatGPT – has been described as the “iPhone moment” in the history of AI. In particular, it has huge implications for AI in healthcare.
As Kimberly Powell, VP and GM for healthcare at Nvidia, recently told me, generative AI provides an answer to the question, “How do you take every medical paper that’s been written, every doctor’s note, and put that into these AI models in a way that you can query that information?”
Generative AI technology has the potential not only to play a role in the discovery of new drugs and personalization of healthcare provision but also in selecting subjects for clinical trials and understanding the impact that these sorts of changes to healthcare are having in the real world.
For example, due to their nature as a language-oriented technology that understands the characteristics of any language, large language models like those powering ChatGPT can be used to understand and speak in the language of nature itself. This includes the language of amino acids that make up DNA codes and the chemical language of SMILES.
Powell told me, “We’re making it more and more accessible for biologists who maybe aren’t trained completely in computer science to be able to train these models and or use these models as part of their everyday work. And we’re seeing an absolute explosion in the number of generative models across biology.”
More than in perhaps any other field, the use of AI in healthcare needs to be backed up by stringent adherence to frameworks around ethics.
While biased data can impact research and decision-making in any industry, it is particularly problematic in healthcare. For example, failing to properly balance training data could result in failure to diagnose or misdiagnosis of illness in under-represented demographic groups.
An uneven distribution of access to this technology could also serve to further widen the disparity in factors such as healthcare outcomes and life expectancy between rich and poor or advanced and developing nations.
And concerns around data privacy and security are clearly paramount when we’re talking about the use of extremely sensitive and valuable information such as patient records and genetic screening.
There’s also the need to ensure AI systems used in healthcare are sufficiently transparent and explainable so that doctors and patients will understand their advice and predictions. Otherwise, it will be difficult for them to trust it!
The Road Ahead
If we can manage to satisfactorily face up to these challenges, then I believe it’s likely that AI will continue to have a revolutionary impact on healthcare provision around the world.
With caution and forethought, AI – and in particular, the advanced generative models we’re seeing emerge today – will help deliver better patient outcomes as well as more efficient delivery of health services.
It will also raise the curtain on a new era of drug discovery and scientific progress while taking us closer to the goals of personalized medicine and curing some of the most devastating illnesses.