Introducing ELLIPS: An Ethics-Centered Approach to Research on LLM-Based Inference of Psychiatric Conditions
Roberta Rocca, Giada Pistilli, Kritika Maheshwari, Riccardo Fusaroli
TL;DR
This paper addresses the ethical gaps in research on inferring neuropsychiatric traits from language by proposing ELLIPS, an ethics-centered toolkit that translates seven core principles into concrete, stage-specific questions guiding target variable selection, data handling, model design, evaluation, sharing, and deployment. Grounded in biomedical and ML ethics, the framework emphasizes autonomy, beneficence, justice, responsible inference, credit allocation, transparency, and social responsibility, and demonstrates its utility with a case study on autism language markers. The authors argue that aligning model development with clinical goals through structured ethical reflection can enhance real-world applicability, safety, and trust, reducing risks of harm and dual-use while promoting equitable, clinician-supported decision-making. Overall, ELLIPS offers a pragmatic pathway to develop LLM-based inference systems that meaningfully improve mental health assessment and remote monitoring while safeguarding patient rights and societal impact. The case study illustrates practical considerations and stakeholder engagement, reinforcing the toolkit’s relevance across NLP and ML for health-related applications.
Abstract
As mental health care systems worldwide struggle to meet demand, there is increasing focus on using language models to infer neuropsychiatric conditions or psychopathological traits from language production. Yet, so far, this research has only delivered solutions with limited clinical applicability, due to insufficient consideration of ethical questions crucial to ensuring the synergy between possible applications and model design. To accelerate progress towards clinically applicable models, our paper charts the ethical landscape of research on language-based inference of psychopathology and provides a practical tool for researchers to navigate it. We identify seven core ethical principles that should guide model development and deployment in this domain, translate them into ELLIPS, an ethical toolkit operationalizing these principles into questions that can guide researchers' choices with respect to data selection, architectures, evaluation, and model deployment, and provide a case study exemplifying its use. With this, we aim to facilitate the emergence of model technology with concrete potential for real-world applicability.