Contact us:
040 4016 5703 099 6344 0404
Follow us:

Implementing Change Part Three: How To Move AI Forward In Healthcare

I’ve explained some of the many ways artificial intelligence could vastly improve healthcare at all levels, from telehealth to the emergency department, from diagnosis to claims processing. I’ve also called out the inherent limits on what AI can do so that we don’t waste time and resources or endanger patients by trying to make it do what it can’t. With both opportunities and limits in mind, what can we do to move AI forward in healthcare?

“AI” needs a new name.

To start, I think we need a new term for machine learning and other technologies as applied to healthcare. The words “artificial intelligence” conjure the very image we don’t want: a machine making its own decisions. Studies have shown the best outcomes result from machines and humans working together, with the AI as a sort of physician’s assistant — preternaturally brilliant in some ways, known to be clueless in others.

I’ve suggested calling it “augmented intelligence” instead. (Perhaps it’s time to bring in the naming consultants.) We need patients to feel comfortable with AI and trust it.

Get patients and physicians involved.

Anyone working in any healthcare aspect knows there’s a huge challenge here. The legacy data we need is huge, yet we don’t have a unique national patient identifier or suboptimal interoperability among systems. AI will need these areas addressed to correlate data to deliver beyond-human insights.

We also need much more cooperation among all stakeholders—physicians, investors, technology developers, hospitals, ethicists, employers, health plans, regulators and, most importantly, patients—to break down specific walls so the right entities work with, not against one another.

Healthcare by nature requires strict adherence to privacy and confidentiality, but this can’t be an excuse to keep essential information hidden where it should be shared transparently upon patient consent.

With healthcare providers, we should be transparent and acknowledge that some tasks and even some jobs may no longer be necessary. We can highlight the ways many professionals’ jobs will change and look for ways that people with a passion for care can advance rather than be eliminated.

Robust investment and oversight are needed.

I hardly need to mention that AI in healthcare will need attentive oversight. The U.S. Food and Drug Administration, American Medical Association, specialty societies such as the American Medical Informatics Association, American College of Radiology, American Academy of Dermatology and other committed guardians of integrity and accountability must be intimately involved to ensure privacy and safety standards.

As an innovation and investment hub, Inflect Health proudly pairs health-tech innovators with those with ongoing front-line medical experience. But we’d like to see other judicious investors nurture AI solutions for patients, physicians, other healthcare providers and payers. The field seems ripe for government investment, too. Just as federal and state programs have fostered more medical degrees for sorely needed roles, a wise investment could deliver better outcomes for more patients.

Frame the problem, then feed AI robust data.

After ensuring appropriate consent, oversight, standards, transparency and physician/patient and other stakeholder engagement, data can be viewed as ready for AI input. Next, as we discussed in my last article, humans should identify the healthcare problem(s) to solve. They should frame the problem wisely and determine constraints and guardrails. Then, they can feed AI with robust data and let it work with physician oversight. It’s long, hard work, but it could bring transformational changes in healthcare.

It’s about progress, not perfection.

The biggest hurdle I see for healthcare AI is that we tend to expect and demand perfection from technology. Let’s be realistic: The current state of healthcare has many, many problems. As we invest, build, test and deploy hopeful solutions, let’s avoid perfectionism and eye a more human goal: How can AI make healthcare better than it is today?

No Comments Yet.

Leave a reply