Beyond the Hype: Real-World AI Use Cases in Healthcare
When we discuss AI, the image that usually springs to mind is a sophisticated, futuristic application that’s entirely novel. Perhaps it involves face detection, a feature not deemed essential in the past, as humans have naturally honed their face-detecting abilities over centuries of evolution. Alternatively, it might involve asking an LLM to undertake creative tasks such as graphic design, which have traditionally been carried out by humans. But have you ever been struck by how humans can be ineffective at certain tasks? We are so used to the ways we do things that it’s rare for us to consider how they could be done more efficiently. I’m not just talking about speed here, but predominantly about the quality of the results we get. Today, I’m going to discuss basic tasks in healthcare and how a practical AI could do them much better than we currently manage. As a society, we urgently need better healthcare. AI could indeed enable new and futuristic applications, but many tasks are already performed by humans with inconsistent quality and standards that are challenging to measure and maintain due to numerous complexities.
The Potential of AI in Healthcare
AI has the potential to significantly enhance experience-based careers involving direct interaction with humans. Practitioners in these fields have to work with humans for a long time before they can establish their own practice. These practitioners heavily rely on historical data, which essentially becomes embedded as experience, to make decisions. As a patient receiving these services, the inconsistency in quality is apparent because it’s virtually impossible to vet all the practitioners. You can’t possibly assess the skills, treatment approach, and expertise of every practitioner, not to mention that you often only have one chance to experience a treatment!
An AI system equipped with comprehensive information and data could generally improve the quality of treatments. Note that I am not referring to life-threatening situations, as our AI systems are not yet equipped to handle such critical tasks. In uncertain scenarios, human judgment, with its speed and moral considerations, is still paramount. Additionally, many of the physical tasks performed by these practitioners can be done with much higher quality by a robot. Humans have remarkable skills and expertise, but they are constrained by the limits of their physical abilities, especially when it comes to intricate and precise tasks. For instance, performing delicate surgeries or manipulating tiny components in medical procedures requires steady hands and utmost precision. However, even the most skilled human hands can have limitations in maintaining consistency and accuracy throughout extended procedures or over their lifetime.
My Experiences
These suggestions come directly from my own journey through the healthcare system. I have dealt directly with dental and eye care, which correlate with the two improvement areas I mentioned earlier. Take a dental procedure, for example. A 3D scan and history of my super-sensitive tooth would provide dentists or robots with more precise guidance. Unfortunately, it’s often challenging to predict which approach a dentist will take for a particular task. Some prefer less invasive procedures, while others lean towards new technologies, leading to variations in outcomes. But with an AI system analyzing data from various procedures and patients, dentists could achieve more consistent and superior results. The same idea applies to retinal nerve fiber layer (RNFL) scans. One eye doctor suggested I could have glaucoma, despite it being highly unlikely at my age. Meanwhile, another doctor declared my RNFL scan normal, attributing the discrepancy to differences in the sample population used for comparison. Putting aside the personal stress and cost of these situations, an AI model, with access to a diverse and comprehensive dataset, could directly address and improve these issues at their roots.
Challenges in Embracing AI-Assisted
Introducing AI into these domains is not without challenges. Firstly, AI deployment often carries the perception of being a high-tech luxury rather than a practical necessity. A perfect example is the Invisalign system, where the underlying technology is often less expensive than the fees doctors charge their patients. Doctors should be at the forefront of embracing this technology, rather than passing on exorbitant costs to patients under the guise of AI usage. While some may have honest concerns about AI’s safety, these concerns should not result in higher costs for patients. It’s essential to shift our perspective and recognize that AI can be a valuable tool in providing standardized and optimal care.
Secondly, it’s crucial to emphasize that AI will not replace doctors; rather, they will remain the primary practitioners. However, for routine and repetitive procedures, even those performed with varying quality in the US, AI can help ensure uniform quality and accuracy. Doctors must understand that AI serves as a valuable assistant in providing standardized and optimal care.
Thirdly, what happens when malpractice occurs? This, I believe, is a major roadblock for AI in healthcare. We can usually chalk up a doctor’s mistake to human error. But with an AI, who’s at fault? It’s tough for any company to survive and avoid lawsuits for minor errors in healthcare. This is a critical issue that simultaneously hinders any company’s ability to scale and enhance their products. As a result, we’re stuck with a decentralized human-based care system with varying quality and inconsistent tools to navigate any issues within this system, like quality measurement and handling malpractice.
Finally, the most substantial challenge might lie in our perspective as patients. Building trust with a competent doctor can be challenging, but once that trust is established, we often rely on their expertise and decisions. How open are we to embracing AI-assisted decisions made by our doctors? Personally, I would trust and even prefer an AI-supported decision from my doctor, but I understand that not everyone may share this perspective.