The world’s leading publication for data science, AI, and ML professionals.

The Safe and Effective Evolution of AI in Healthcare

By: Dan Riskin, M.D., FACS, founder and CEO of Verantos

AI allows us to practice healthcare using the brain of a computer and the heart of a human

Source: Verantos.com
Source: Verantos.com

By: Dan Riskin, M.D., FACS, founder and CEO of Verantos

Early in my career as a trauma surgeon, I saw a patient who was in a horrible car crash and had suffered multiple organ injuries. I had multiple options of how to treat the patient, but limited evidence to make a decision. I thought to myself, if only a computer could look up other individuals who had suffered from a similar pattern of injury and tell me which approach worked best.

That technology didn’t exist at the time. But, if it had, that patient would have benefitted. Instead, his care was dependent on one human, and one human can’t possibly know all the right answers. Today though, this technology does exist. Artificial Intelligence (AI) allows us to practice healthcare using the brain of a computer and the heart of a human.

AI as an Appliance

We often see horror films portraying artificial intelligence (AI) in the form of robots set on the destruction of humanity. This line of thought, which centers around machines eventually replacing humans, is neither constructive nor true. I like to think of AI as a needed support. Decades ago, people washed their laundry and dishes manually. Eventually we found that machines could handle these tasks more effectively and efficiently. The same is true of AI. As we find better solutions for aspects of healthcare that are repetitive or require an inhuman memory, there isn’t a doubt in my mind that we should benefit from machines. It’s not about replacing humans, rather AI supplements what humans can do.

Source: Verantos.com
Source: Verantos.com

Solving Bias in AI

The issues surrounding bias in AI have been discussed at length, and for good reason. AI algorithms are highly dependent upon source data, which is often based on trials implemented at advanced and expensive health systems. The individuals who have access to these health systems often fit a specific economic, social, and racial profile. This means that the most available source data stems from wealthy and white populations.

Verantos addresses these inequities by using precision medicine to identify preferred therapies for subgroups, including underrepresented populations. We make it possible for AI to analyze millions of pages of de-identified health records to determine what is and isn’t working for these subgroups. After all, the best way to avoid bias in AI is to use well-distributed source data.

Conclusion

We are at a unique and exciting time in the healthcare industry. With the confluence of enough data, robust AI and sufficient compute power, we are able to aspire to the next frontier of healthcare – a learning health system that tailors therapy to each individual based on what was learned from past care for thousands, or even millions, of similar individuals.

Dan Riskin, MD, MBA is Founder and CEO of Verantos, the only RWE-specialized firm that runs regulatory-grade EHR-based studies. He is Adjunct Professor of Surgery and Adjunct Professor of Biomedical Informatics Research at Stanford University. Dan is an expert in healthcare AI and has developed products that influence the care of millions of patients annually.


Written By

Topics:

Related Articles