John Pestian, PhD, MBA, an Endowed Professor at Cincinnati Children’s Hospital Medical Center and the University of Cincinnati College of Medicine, recently presented findings from the Decode mental health initiative at the AI4Peace Symposium in Rome. This event, held at the Pontifical Lateran University with support from the Vatican's Pontifical Academy for Life, attracted over 250 attendees from various countries to explore the scientific and ethical implications of artificial intelligence.

Read More

Pope Leo XIV conveyed a message to the participants, expressing hope that their work would foster a commitment to using technology for the common good. This theme of building AI that respects human dignity was central to the discussions, framing the significance of the Decode program's developments.

To effectively assist individuals with mental health issues, understanding both biological and cognitive factors is crucial. Traditionally, medicine employs biomarkers such as blood glucose levels for conditions like diabetes to guide treatment. In contrast, mental health care lacks such objective measures, a gap that Decode aims to address. The program seeks to identify significant patterns in language that can help clinicians tailor care to individuals.

Pestian shared insights regarding a study examining the language used by individuals in crisis, including narratives from English, French, and Chinese speakers. The research revealed cultural differences and variations within cultures that are critical for understanding a person’s unique experience. AI, while adept at identifying patterns in vast datasets, can inadvertently flatten these nuances, which could erase vital information necessary for effective crisis intervention.

Decode employs mathematical geometry to analyze language structure, allowing clinicians to visualize distress beyond what biological metrics can show. This approach assesses the relationships among cognitive markers in language, providing a richer understanding of an individual’s mental state.

By integrating thought markers with traditional biomarkers, clinicians can gain a comprehensive view of their patients. Pestian likened this process to mapping a landscape, where the intricate details—both the words used and their interrelations—shape the overall understanding of an individual's experiences.

A notable experiment involved analyzing language samples from 150 people in crisis, many of whom faced suicide risks. Initially, an AI model provided similar interpretations for 62% of individuals, regardless of their diverse backgrounds. This highlights a significant issue in mental health care, as critical distinctions among individual experiences were overlooked.

To rectify this, researchers adapted the AI model to maintain meaningful differences, ensuring it recognized both cultural and individual variations during analysis. This adjustment allows for a more tailored approach to mental health interventions, emphasizing the importance of individualized care amid cultural contexts.

This research is particularly relevant for children and families in both specialized centers like Cincinnati Children’s and underserved communities with limited access to mental health services. The tools developed through Decode aim to assist clinicians in recognizing unique expressions of distress in children, ensuring that technology enhances understanding rather than oversimplifies it.

Pestian concluded his reflections by recalling the Pope's message on the ethical responsibility of those developing AI. He emphasized that when interpreting experiences, AI must recognize the distinctiveness of each individual. As the Decode team continues their work, the focus remains on fostering AI systems that honor personal narratives and serve the common good in mental health care.

For anyone experiencing a crisis, contacting the 988 Suicide & Crisis Lifeline is encouraged.