This article is based on a presentation at IBA’s 2023 Spring Conference on Healthcare Analytics by Sagar Samtani, Assistant Professor of Operations and Decision Technologies, Weimer Faculty Fellow and Director of Kelley’s Data Science and Artificial Intelligence Lab (DSAIL).
Sagar Samtani admitted during his presentation during the Spring 2023 Institute of Business Analytics Conference on Healthcare Analytics, “From Subjective to Objective Measurements of Mental Health: An Artificial Intelligence (AI)-enabled Analytics Perspective,” that the audience may be skeptical why, as a cybersecurity expert, he was talking about mental health.
“Shouldn’t I be in a basement, hacking something?” Samtani joked.
Samtani, an Assistant Professor of Operations and Decision Technologies at the Kelley School of Business, has a deep knowledge of cybersecurity, though. He completed his doctoral studies at the University of Arizona, where he was a part of the AI lab and his work focused on developing machine learning and deep learning-based approaches, specifically in cybersecurity.
At IU, Samtani runs a research lab, Kelley’s Data Science and Artificial Intelligence Lab (DSAIL), focused on developing AI-enabled analytics techniques for different application areas, primarily in cybersecurity and mental health. His interest in mental health started while he was working on a workforce development grant as a doctoral student.
“I noticed that a lot of people who are in the information security or cybersecurity space left their jobs after a year or two,” Samtani said. “There’s a lot of turnover, which becomes a workforce retention issue, not a workforce development issue, and one of the major reasons that’s been cited for this significant turnover is actually mental health.”
Samtani mentioned a few causes of mental health concerns for cybersecurity professionals, including being slow to be praised, having to work night shifts, lacking real time off, and being “equipped to be blamed” when something goes wrong.
“You don’t know that a lot of cybersecurity professionals even exist until something goes wrong, then it’s like, what happened?” Samtani said. “We should have defended against this.”
There are ways for Information Technology (IT) leaders to address the growing mental health crisis, Samtani said, including sharing resources to cope with stress and evaluating the impact of night shifts.
Process for Recognizing Mental Health Concerns
In addition to these specific tips, Samtani shared a process for recognizing and addressing mental health concerns with three major steps — and the limitations around these steps.
- Identify Behaviors: A supervisor or manager can look at the individuals who they manage and try to determine if their behaviors have changed. The higher-up is likely not a trained therapist, though, so it is unfair to expect them to be able to recognize these behaviors.
- Recommend Services: Based on that judgment or assessment, the supervisor may recommend certain types of downstream services, but it takes a lot of time and knowledge to be able to recommend these services.
- Execute interventions: The individual participates in one of the downstream intervention services. However, the providers at these services are often overworked, which can lead to inconsistency in treatment, Samtani said.
Using AI to Identify Mental Health Concerns
This part of the presentation is where AI comes into play, Samtani said.
“I always coach my students to ask, ‘how do you translate real problems and think about the data that’s necessary to solve those problems? How do you develop technical solutions around that that could reasonably address those issues?” Samtani said.
He offered possible AI solutions for each step of the previously mentioned three-part process: sensor signal-based approaches for identifying behaviors; using recommender systems and semantic matching to recommend services; and training chatbots or agents to provide, at some level, therapeutic or mental health services.
Samtani focused on AI solutions for the first step in the process, starting with improving mental health literacy, or knowledge about mental disorders that aid their recognition and management, to better recognize behaviors. AI-enabled analytics techniques aim to move from subjective measurements of these behaviors to objective measurements, Samtani explained.
“Sensor signal data can capture multiple dimensions of human behavior, such as physical, social, and sleep, all of which we know lead to certain types of mental health conditions,” Samtani said. “It’s often very high velocity, though, which makes manual analysis very difficult to do.”
Samtani’s lab is now researching how to analyze this data to identify individuals with depressive behaviors. For major depressive episodes, specific sensors that are correlated with predicting depressive output include how much time a phone screen is locked, which means a person may be sleeping more or less; and data from the accelerometer or microphone, which identify noise made, steps taken, and conversations.
If behaviors can be identified objectively, he said, they can trigger and recommend specific interventions.
“We want to make a positive impact on the IU community,” Samtani said. “This model could help students gain higher self awareness to seek out professional help while feeling depressed, and improve mental health literacy to help people with warning signs of depressive behaviors.”
Leave a Reply