NeuroFlow NLP Technology Surfaces Suicide Risk That Assessments Alone May Miss

We’re excited to introduce “Briefings,” a special series of our Bridging the Gap podcast. These short, news-focused episodes hit on timely trends that are impacting healthcare. In our inaugural episode, NeuroFlow Chief Product Officer Julia Kastner explored new research on NeuroFlow’s natural language processing (NLP) technology and how it’s surfacing suicidal ideation when traditional screening falls short. 

This episode highlights NLP’s potential to revolutionize the way suicidal ideation is detected and addressed. Read on for top sound bites and tune into the full episode below!

How NLP Became a Part of NeuroFlow’s Platform

One of the popular features of NeuroFlow’s patient engagement platform is journaling. It offers an opportunity for individuals to reflect and creates rich data sets that can indicate changes in wellbeing. “We originally heard from some of our clinical users that they were having trouble reading all of the journals in our mobile application,” says Kastner, “So we thought, couldn’t we use something like natural language processing to read the journals on behalf of the clinicians? . . . In particular, they were concerned that some of those journals had mentions of suicidality.”

This led to the creation of an NLP algorithm capable of discerning genuine suicidal ideation expressions from benign statements, ensuring that organizations receive pertinent alerts without being overwhelmed by false alarms. This technological advancement addresses a critical need within the behavioral health sector—the capacity to detect and act upon signs of distress that may otherwise go unnoticed.

New Research Demonstrates NLP’s Ability to Detect Suicidal Ideation

New research, published in Innovations in Digital Health, Diagnostics, and Biomarkers, analyzed NeuroFlow’s NLP-created alerts that indicated suicidal ideation. “We identified 425 individuals who were flagged for suicidal ideation by our NLP technology,” says Kastner, “We looked to see if they had done a PHQ-9 in the past 30 days, and what we found was that 58% of the people that we looked at who had triggered our NLP algorithm either didn’t have a PHQ-9 at all, or had not identified self-harm in the last question. That means that 58% of people that we found were people with suicidality that we wouldn’t have seen through traditional clinical methods.”

Kastner adds, “Without this technology, a lot of risk would’ve been missed. . . We’re really excited to use this as an example of what technology can do when you pair AI and analysis with a human in the loop.”

Looking Ahead

The ability to uncover hidden behavioral health risks epitomizes the power of combining AI with human intervention, creating a safety net that catches individuals falling through the cracks. 

“We’re really excited to think about the possibilities of natural language processing as we’ve been using it, but also the potential of large language models and other types of data science and data analysis to look at large amounts of data in the healthcare ecosystem,” says Kastner. “There is signal and EHR data, there is signal in claims data, there is signal in pharmacy data. There’s a lot of opportunities to look for those changes in behavioral health needs either retrospectively or in real time, and those are the initiatives our team is working on today.”

Ultimately, through surfacing risk earlier, Kastner argues that healthcare leaders can improve access to care. “My general philosophy is that innovation in this space is providing better access,” says Kastner. “The best way to increase access to services like behavioral health is to find the people that need it most and pair them with the right level of care.” 

Ellen Harvey is the Senior Content Marketing Manager at NeuroFlow. She has over nine years of experience writing about technology and innovation for business leaders. At NeuroFlow, she writes about prominent trends in behavioral health and illustrates how NeuroFlow's technology helps healthcare, payor, and government organizations improve the well-being of their constituents.

Enter your keyword