ImpuLS-AI (Early and low-cost diagnosis of impulse control disorders in children and adolescents based on automated analysis of language and sleep patterns) is a joint research project between the HTWG Konstanz Ubiquitous Computing Lab (UC-Lab) and the Computational Sociology research group (GIADSc) at Universidad Tecnológica de Pereira. The project addresses an urgent challenge: the marked increase in mental health problems among children and adolescents, especially since the COVID-19 pandemic, and the need for scalable, accessible, and cost-effective approaches that can support early detection and timely intervention.

At its core, ImpuLS-AI aims to develop a methodology and a practical toolchain that combines automated speech/language analysis with sleep and stress pattern assessment—collected via mobile and low-barrier measurement setups—to support the early identification of impulse control disorder (ICD)-related symptoms in young people. The project is motivated by evidence that stress and sleep changes are tightly intertwined with mental health in youth, and that the pandemic has intensified vulnerability to both internalizing (e.g., anxiety, depression) and externalizing symptoms (e.g., hyperactivity, inattention, behavioral problems including ICD-related patterns).

The scientific ambition of ImpuLS-AI is framed by two guiding hypotheses. First, the project assumes that applying AI techniques to speech can reveal structural and semantic patterns associated with ICD—patterns that may be difficult to capture consistently through traditional clinical observation alone. Second, it posits that there are meaningful correlations between stress characteristics, sleep features, and ICD-related symptom profiles, and that these correlations can be leveraged to design predictive and diagnostic algorithms that are suitable for youth and feasible in real-world settings. Importantly, the project does not treat sleep and stress as secondary signals; instead, it investigates them as potentially informative indicators that can strengthen early assessment and personalize follow-up actions.

A distinctive feature of ImpuLS-AI is its integrative approach: it combines (1) objective digital signals (sleep and stress-related measures) with (2) AI-based language assessment, and (3) grounds the overall methodology in validated clinical/psychometric perspectives. This combination is intended to reduce reliance on single-source assessments and to improve robustness—especially when working with children and adolescents, where symptom expression can vary by context (school, family, peer environment) and where early support can have long-term impact. 

Overall, ImpuLS-AI seeks to deliver new scientific insights into the interplay of stress, sleep, language patterns, and impulse control in youth, and to translate these insights into practical, low-cost, technology-supported assessment components. The results are intended to be transferable across contexts—supporting innovation in mental health technology in Colombia while remaining relevant to similar challenges in Germany—thereby strengthening digital transformation in healthcare through evidence-based, scalable tools.