To Help Achieve Life-Saving Discoveries, AI Must Access ‘Underutilized, Undervalued’ Findings

Of all the ways in which AI can transform healthcare, one of the most promising is in revolutionizing research. These new technologies can empower every part of the sector to understand patients and providers like never before.

It’s not just that AI-powered tools can collect more information about more patients than ever before. It’s also that they can unlock what has long been a largely untapped resource: qualitative data.

“Public health research that addresses chronic disease has historically underutilized and undervalued qualitative methods,” a study in the Annual Review of Public Health explained. This has “limited the field’s ability” to get a more in-depth understanding of health behaviors; determine why and how a remedy did or did not work; and test out new theories, the study added.

There are numerous reasons this has happened. Quantitative data such as numerical, multiple-choice, and “yes or no” questions can seem like a more concrete basis for decisions. Reading through freeform responses to open-ended questions can be tedious. And even when researchers piece through those responses, their meanings can be nebulous.

But now, healthcare companies can collect insights from all of that unused qualitative data at scale. AI-powered platforms with more advanced approaches to natural language processing (NLP), trained in the specific vocabulary of any scientific topic, can read all of those responses.

These platforms can spot trends, common problems, areas of confusion and more. And they can provide summaries so that researchers, providers, payers and other stakeholders get crucial insights at a glance.

The best new systems pull qualitative and quantitative data together, offering the best of both. On their own, qualitative data isn’t clearly representative, while quantitative data lacks the nuance and color necessary to understand the results. When an AI tool unifies them, it can provide three-dimensional findings. The tool can also recommend next steps for what to research, test, or survey, which populations to focus on, and more.

All this is just the beginning. The scientific community is inundated with new studies on a daily basis, including many with overlapping themes. AI-powered platforms can collect data sets from disparate sources and check them for quality, duplication, relevance, and more. These platforms can then create even more extensive, understandable findings to help healthcare professionals make decisions.

They can segment the conclusions based on any number of traits. So, for example, a healthcare professional can describe a specific patient and get instant feedback highlighting the most pertinent results.

Strengthening digital twins

All of those abilities pave the way for new and better digital twins — virtual representations of real people that are more representative. Outside of the healthcare space, these are increasingly used to model human behavior and decisions. And digital twins are being used for things like clinical trials. But these “twins” are not all made equal.

The most useful ones are as rich in detail as possible, based on robust collections of information about real people. Creating these kinds of digital twins requires a deep foundation of both qualitative and quantitative data, which must be updated in real time as new information is collected about patients in the real world.

With high-quality digital twins available, healthcare companies open up a world of potential. They can ask questions that are usually off the table due to privacy issues. They can test multiple therapies, medicines, and other remedies simultaneously. These twins can also be designed to meet a unique combination of characteristics at any given time, including age, medical history, allergies, environmental factors, social determinants, and more.

None of this means risking anyone’s health. As with all research, digital twins cannot definitively demonstrate how any individual will respond. Real testing, with real people, is of course as necessary as ever.

But when these AI tools are “fed” all the information about any given drug or therapy and tasked with exploring how the digital twins respond, they can discover important things — benefits, complications, adverse reactions, risk factors and more. When they’re built with both quantitative and qualitative data, they do the job much more effectively.

There are all sorts of use cases for digital twins in healthcare spaces. Pharmaceutical companies can learn about perceptions of drugs and vaccines, as well as barriers to patient or doctor acceptance of new treatments, and test out entirely new ways to market them. Providers, including practices and hospitals, can use these for brand tracking. Public health agencies can use them to help design initiatives that are most likely to succeed.

In every case, the technology used will rely on having the best possible collection of information. Even the most expensive, complex systems are limited by the data they’re given access to. So as medical and healthcare organizations look for ways to move forward, qualitative data should serve as a linchpin. People aren’t numbers, and any one individual’s descriptions, thoughts and feelings don’t represent the masses. But when you bring all those numbers and descriptions together, you’ve got a much greater likelihood of success — helping to improve, and even save, lives.

Photo: MirageC, Getty Images

This post appears through the MedCity Influencers program. Anyone can publish their perspective on business and innovation in healthcare on MedCity News through MedCity Influencers. Click here to find out how.