While many people, especially healthcare insiders, understand the manifold benefits of artificial intelligence (AI), it’s gotten a bad rep over the years. Fears of machines taking over our relationships and jobs—plus the scariness of their access to our personal data and life choices—have kept many people and industries hesitant to fully embrace the increasingly promising technology. And many healthcare providers have their own particular hang-ups.
Take, for example, dermatologists. Many of these specialists prefer in-office visits—as opposed to AI-powered, remote analyses of patients’ skin conditions—to ensure that diagnosis and treatment monitoring is something they can do with their own trained eyes. The pandemic forced a culture change, however, with telemedicine becoming the norm. Consequently, dermatologists have had to rely on whatever HIPAA-compliant video or photo methods they could secure to stay connected to their patients remotely, especially those with chronic skin diseases.
Enter AI, which has been a developmental focus in dermatology for some time, albeit with slow adoption. While some doctors have for a long time resisted integrating AI algorithms into their practices, more and more are exploring the benefits of remote care to augment rather than replace their practice. It started with some dermatologists seeing AI’s power to facilitate diagnosis.
But over the course of the pandemic, with physicians having to go months and months without seeing certain patients in some cases, more providers are seeing the promise of AI for much more. They’ve come to appreciate the ability of machine learning to not only identify skin concerns, sometimes earlier—AI can correctly identify 95% of melanomas—but also closely track the progress of chronic diseases in an easy and regular way. What’s more, the constant monitoring afforded by AI often yields a level of precision that doctors could never achieve over the course of just a handful of office visits.
According to a study in The Lancet Oncology, even though AI algorithms can actually outperform humans in some areas of dermatology, certain limitations mean that they cannot take doctors’ place. For example, AI tends to underperform when it comes to images that are out of distribution, a machine-learning term for anomalous inputs.
The "new normal" of our post-pandemic world has now established telemedicine and remote care with a secure role in a patient's journey. And for all of us, both those with current skin concerns and concerns that we haven't seen yet, AI is a tool we can use to play an active role in our health and stay more closely connected to our physicians. Dermatology needs AI, and we do, too.
