Detecting Loneliness in Call Centre Interactions
Queensland University of Technology
The Challenge
From community care and social services to aged care, housing, and mental health referrals, contact centres often provide frontline support in emotionally complex situations where callers are frequently vulnerable, and the quality of interaction can directly affect their wellbeing. Trust in how data is used to inform their experience is particularly important.
- Clients whose conversations are not typically designed to recognise or respond to distress, anxiety, or psychosocial signals.
- Frontline agents who lack the experience to meaningfully respond when clients disclose sensitive data. This can lead to moral fatigue, performance issues, and high turnover.
- Service providers who engage in repeated contact, miss opportunities for early intervention of at-risk clients, or suffer from reduced service quality.
The Partnership
The partnership involves close collaboration to detect signs of loneliness both qualitatively and quantitatively through large-scale data analysis, observations, and coding of past calls. Researchers work directly with call centre staff and management to design operational changes. Through co-development workshops, they establish frontline protocols analogous to domestic violence response procedures.
The Solution
There is a growing opportunity to deploy AI and digital tools to better support both staff and clients – especially in detecting subtle psychosocial cues and prompting more human-centred, responsive interactions at scale. This approach demonstrates how digitalisation, while often reducing face-to-face contact, can lead to more meaningful conversations precisely where they matter most. It provides organisations with a blueprint for how to digitalise responsibly in healthcare and beyond.
Recognition & Impact
Project Highlights
-
Launch of loneliness-detection whitepaper report to showcase first results.
Ongoing collaboration to:
- Develop a psychosocial signal detection model, using real-world call data to identify subtle indicators of distress, anxiety, or other unmet social and emotional needs.
- Co-design interventions with psychologists that empower call centre agents engage with clients –including suggested prompts, phrasing, contextual cues and tone adjustments.
- Examine trust boundaries by testing which types of AI-augmented interactions are perceived by clients as helpful versus intrusive. The project will explore how perceptions shift depending on context, delivery, and the degree of benefit to the client.
- Evaluate impact on service outcomes (e.g. call resolution, repeat contact, agent confidence), operational metrics (e.g., handling times), and broader wellbeing metrics for both clients and staff.
Contact Person & Details
- Dr Nadine Ostern, NIIN Research Chair in Trusted Retail, Centre for Future Enterprise, Queensland University of Technology