A lot of new artificial intelligence tools are being deployed in senior care settings, with one main goal being the assumption of administrative tasks to lessen the mental strain on staff members
Now, a new study is looking at whether AI can directly detect levels of anxiety and depression in caregivers.
The AI-evaluated transcripts of psychiatric appointments of more than 800 healthcare workers, roughly half of them nurses, and found a higher risk for anxiety and depression among this cohort, study authors reported.
AI also was able to categorize employees’ concerns into useful groupings, such as whether their fears were related to viruses such as COVID-19, or to working with patients.
Although the psych consultations took place during the pandemic period, senior living remains a high-risk environment for viral breakouts, either for a new COVID strain or the flu. Additionally, caregivers at skilled nursing facilities, the same as in hospitals, may work long hours while tending to many patients.
High levels of mental staff burnout are a major challenge in healthcare, perhaps worst of all in senior living. Within the last year, 96% of senior living communities said they were facing a staffing shortage; at one point, there was an unfilled need for 400,000 senior caregivers overall, McKnight’s reported earlier this year.
The researchers touted the AI’s ability to identify specific mental health concerns via language patterns, and in a manner less intrusive than a formal survey.
“Unlike traditional assessment methods, such as surveys and interviews, NLP [natural language processing] markers from psychotherapy platforms present a passive and less burdensome way to assess therapy-seekers’ mental health,” the study authors explained.
“It’s akin to the digital biomarkers developed from wearable and smartphone data. Defining and validating markers of anxiety and depression could lead to more accurate and reliable assessments, which would be beneficial for health care providers. ”The study was led by a research team at NYU and was published recently in the Journal of Medical Internet Research AI.
