The Future of Artificial Intelligence in Health Care

The use of artificial intelligence (AI) technologies across industries, including health care, has existed in one form or another for decades. But now more than ever, there is unparalleled promise for what AI in health care can achieve in saving and improving lives.

Advances in machine learning (ML), deep learning (DL), and natural language processing (NLP) in health care have been instrumental in gaining efficiencies at every level, from billing to imaging to diagnosis; delivering better, more decentralized care; preventing illness and disease; lessening health professionals’ workloads; and, crucially, lowering costs for patients.

While dystopian visions of robots performing surgeries or replacing doctors and nurses altogether may be common tropes in popular culture, the real issues data scientists, health professionals, and patients face are how and when organizations roadmap the adoption and exploitation of AI technologies in a way that is not only useful to humans and technologically viable, but socially responsible and sustainable.

Below, we cover areas of health care that are currently being improved upon through AI, as well as areas that are ripe for further disruption and innovation.

Before the global pandemic, telehealth visits were practically nonexistent, but this wasn’t for lack of technology. Videoconferencing, digital charts, voice recording and recognition, and appointment-setting software already existed. Telehealth was not used because medical insurance providers did not recognize a virtual patient-provider visit as a qualified, insurance-covered event.

All that changed in an instant. The pandemic brought to the forefront crises in public health, access to care, and skyrocketing costs—none of which can or should be ignored—but it also brought the normalization of virtual health care visits, which were long overdue, positive, and necessary. This was the result of a forced rethinking of the entire structure of the delivery of care, and it signaled the urgent need to decentralize services and empower patients with access and information, and ultimately, prevent and reduce hospitalizations.

In that sense, technological capabilities alone weren’t enough to shift an entrenched system of operation. When it finally happened, eradicating physical distance from a medical provider as a barrier to care was a hugely transformative event.

Distinct and inarguable progress has been made with NLP systems and how these systems are now capable of analyzing speech, understanding language, and even producing translation. Clinical documentation can now be understood and analyzed for insights in a way it hadn’t before.

But NLP systems are still a long way from where they could be. They still struggle enormously with recognizing accented speech, for example, and with providing simultaneous translation, which means that at the patient-provider level there is still a substantial—and difficult to meet—need for multilingual nurses and doctors in hospitals and urgent care centers and human translators to accompany patients on appointments, so that the quality of communication does not disrupt diagnosis or treatment.

Once it is optimized and adopted by the health care industry, accurate simultaneous voice translation by an application will change lives in a powerful way.

Machine learning is slowly replacing rule-based systems in health care organizations with approaches based on interpreting data using proprietary medical algorithms.¹ Using large samples of data to recognize patterns, ML has already helped health professionals and patients by automating medical billing and providing clinical support.

Beyond that, ML can help evolve the jobs of doctors and radiologists—not replace them—through algorithmic support of data, helping these health professionals make intelligent decisions by processing vast amounts of data at an incredible speed and finding patterns that were previously undetected. For example, in rural areas ML-driven tools can help create the same level of precision diagnosis for certain types of medical events, such as a stroke, that would otherwise have to be made at a completely different geographic location by a specialist.

When it comes to how a person responds to medical treatment, research has shown that humans have vastly different reactions—physiological, biological, and behavioral—and this creates a need for the kind of treatment that is tailored uniquely and specifically.

Given how important data-intensive assays are to revealing appropriate intervention targets and strategies for personalizing medicines, AI can play an important role in the development of personalized medicines at all relevant phases of the clinical development and implementation of new personalized health products, from finding appropriate intervention targets to testing them for their utility.²

Again, the most important challenge in moving personalized medicine forward lies in effectively refining and integrating a vast amount of data into a tailored treatment plan. In the future, this will mean significant time and effort saved in patient treatment trial and error, even when medical history is known, thus enabling better prevention, faster diagnoses, improved outcomes, fewer hospitalizations, and a healthier population.

“The common misconception that AI will somehow substitute humans misses on a big point,” says Eric Nesser, a Chicago-based technologist and entrepreneur. “AI is more human than we think. It cannot exist without data, and that data is provided by us.”

It is important to remember that AI is first and foremost a means to an end, a way to solve problems alongside human effort. AI truly is only as useful, thoughtful, unbiased, and sophisticated as the humans behind it.

¹https://www.foreseemed.com/artificial-intelligence-in-healthcarearrow_upward
²Nicholas J. Schork, PhD “AI and Personalized Medicine”arrow_upwardReturn to footnote reference

Return to footnote reference