Natural language processing (NLP) is the ability of computers to understand human speech terms and written text. Within healthcare, the adoption of NLP is now rising because of its recognized potential to search, analyze and interpret large amounts of patient data. NLP technology has the ability to harness relevant insights and concepts from data that is buried in text form, which otherwise would typically take clinicians / physicians several hours to read through and comprehend.
NLP in healthcare can accurately give voice to the massive amount of unstructured data of the healthcare universe, and here are the three reasons why I believe 2023 is the year that we will see increased adoption of NLP across the industry.
1.The kinks have been ironed out
First, let’s recognize that NLP can be applied in several contexts. It can refer to voice-to-text recognition. It can also be used for handwriting recognition. It can be used for language generation, as many of us have seen with recent news about OpenAI’s ChatGPT. But in the context of this article, we are using NLP for content intelligence – or information extraction – of the written word.
About five years ago, machine learning technology achieved a great milestone: it became possible to cost effectively train algorithms with massive amounts of data. That innovation enabled NLP for content intelligence. Machine learning was beginning to be applied to massive amounts of narrative data to build NLP models that could identify key concepts described in text.
And more recently, because the cost to develop a model has dropped, it has become economically feasible to develop industry-specific models.
For example, in the legal industry, NLP has been used for e-discovery. Lawyers use NLP to mine documentation delivered during the discovery phase to make it easier to find critical information buried in the content. And there has been progress more recently in leveraging NLP in healthcare – behavioral health and health and human services more specifically.
Initial content intelligence efforts in health and human services were typically custom projects that were meant to analyze data at a specific point in time, rather than providing a tool that could be accessed on a daily basis. The expertise and effort necessary to “teach” deep healthcare context was too burdensome for many and resulted in project failure – or never getting started at all.
In the last year or so, industry-specific solutions have become commercially available, because the pilots to prove them out have completed. These pilots benefited from the collaboration between data scientists and customers/users who refined the language model for that industry’s need.
So, the kinks have been ironed out. The technology is mature and stable, innovative tech companies have built easily obtainable mission-specific SaaS solutions with deep context, and customers are now reaping the rewards.
- The value has been proven
The direct ROI achieved by organizations leveraging NLP has been realized.
The health and human services industry in particular has seen great results. Several organizations want to help case workers quickly and easily access better insights that would paint a picture of a whole case, without having to spend hours of time flipping through notes. The ability to quickly find the right information to provide a full picture of the case helps particularly with case transfers, so the new frontline worker can quickly understand what has happened. Some caseworkers claim that their NLP platform has saved them five hours per week in administrative tasks. This is significant.
NLP platforms also help organizations have a better understanding of social determinants of health (SDOH). Typically, it would take a careful review of the entire case history to understand topics like history of drug usage or housing insecurity – two SDOH factors that significantly impact overall wellbeing. But with all the color, detail and deeper descriptions living within the unstructured data, an NLP tool enables caseworkers to see early warning signs pertaining to SDOH in real time.
Needless to say, it’s incredibly helpful for families when caseworkers can pull out information such as this from unstructured data earlier in the process.
- The timing is right
Staff shortages and burnouts are currently a real challenge for healthcare organizations. According to a study published in Mayo Clinic Proceedings, the clinician burnout rate among U.S. physicians spiked dramatically during the first two years of the Covid-19 pandemic after six years of decline.
Additional research has shown that 64% of burnout is attributed to administrative burden, which is certainly contributing to caseworkers’ breaking points. With caseworkers stretched so thinly, attrition remains high.
Remember: there is a loss of case knowledge that occurs with attrition, and that loss directly impacts outcomes. When new caregivers are assigned cases, they simply don’t have hours of time to read entire files, which can result in interruptions in the continuum of care, particularly in complex cases.
As a result, caseworkers and clinicians are spending too much time away from the people in their care. They are being pulled beyond their limits. Couple this with the impact on outcomes from lost case knowledge, and it becomes clear that the status quo simply cannot continue if we want to maintain a reliable and smoothly functioning healthcare system.
At the same time, there are significant advances in cost-effective machine learning tools, particularly NLP, that can alleviate much of that stress. The time is right for healthcare providers to lean on available tools. Therefore, I believe 2023 will be the year NLP will take off.
Photo: Peach_iStock, Getty Images