Despite what some of the splashier headlines might shout, artificially intelligent robots are unlikely to replace human doctors, and technology won’t redefine patient experiences anytime soon. However, AI has come to claim an essential place in the healthcare sector in recent years; today, it stands not as a means to replace human work, but as an invaluable support for human success.
In January, Accenture’s senior managing director and head of global health practice, Dr. Kaveh Safavi, pressed this very point in an article for Healthcare IT News.
“In 2019, there was a fundamental shift in how AI is understood,” Dr. Safavi shared. “People started to realize that AI is best used when augmenting the work of humans, rather than substituting them. This will have a tremendous impact on how AI is utilized in 2020, and I see it becoming a self-running engine for growth across healthcare, helping clinicians make better decisions and extending their reach.”
The growth Dr. Safavi refers to is already underway in several sectors. Take the Israel-based startup Sight Diagnostics as an example — over the last few years, the company has developed and begun using a desktop machine that using AI to quickly analyze patient blood samples and perform blood counts in situ. Similar strides have been taken in radiology; just last year, MIT professor Regina Barzilay developed an AI-powered tool that could predict a patient’s risk for breast cancer at equal or better accuracy than conventional methods.
AI has also demonstrated particular efficacy in healthcare administration. According to an article recently published in the Harvard Business Review, the use of AI to help categorize unstructured fax messages eliminated an incredible three million hours of work from the healthcare system in 2017. Clinical note-taking tools such as Suki similarly lessen administrative burdens, allow doctors to spend more time with their patients, and save provider organizations money.
All of these AI-powered tools are undeniably useful for providers. However, the health sector could be doing more to leverage artificial intelligence across the healthcare system as a whole, rather than boosting niche capabilities and managing overflow from our currently overburdened system. After all, understanding patterns is what AI does best — so, why shouldn’t we apply it to improve patient care and experience on a broader scale?
If we leverage it correctly, artificial intelligence can help providers use anonymized patient population data to manage care better overall. Providers can make use of the analytical capabilities that AI offers to better collaborate on a patient’s behalf and avoid potentially conflicting care solutions.
Physicians offer the best care they can for the health concerns under their jurisdiction; however, the treatment plans they create can, in some cases, come into conflict with those developed by other specialists for other conditions. This is a particularly pressing problem for elderly patients who face multiple or chronic conditions.
According to one 2017 study published in Drug Safety, 96 percent of surveyed elderly patients found discrepancies between their medication lists and those maintained by their physicians. Similar — if somewhat less drastic — results were found in an earlier study, with the addition that 51 percent of patients were taking medications that were not reflected in their providers’ records. The drug conflicts that these knowledge gaps create can pose a real danger to patient health outcomes and wellness.
However, AI stands a chance of limiting that risk and vastly improving provider collaboration. In 2019, researchers at Penn State developed an AI-powered tool that can analyze data on drug interactions and potentially warn providers about the risks that certain medication combinations could pose to a patient’s health. The study leveraged data pulled from FDA- and other health organization’s reports to create a potentially invaluable alert system and pre-emptive risk-limiting tool.
“This study is of very high importance,” study leader Soundar Kumara told Health IT Analytics of the matter. “Most patients are not on one single drug. They’re on multiple drugs. A study like this is of immense use to these people.”
In this way, AI could vastly improve the quality of care that patients receive and enhance collaboration between providers — both of which would be useful as the healthcare sector transitions to a value-based, outcomes-centric care model.
However, this idea of improving patient outcomes through AI isn’t limited to preventing drug interactions. On a broader scale, providers could use anonymized, HIPAA-compliant patient data to assess which care solutions are more outcome- and cost-effective than others. This would allow providers to, over time, hone in on treatment plans that are optimized for maximum effectiveness and cost-savings.
Once we achieve that degree of optimization, providers will be able to drive down healthcare costs, improve patient satisfaction, and ensure that the healthcare system as a whole is as efficient and effective as it can possibly be.