Non-attendance at scheduled psychological assessments represents a critical barrier to suicide prevention efforts among adolescents. Early identification of suicide risk is only effective if followed by appropriate clinical evaluation and intervention. However, many adolescents who screen positive for suicide risk fail to attend follow-up assessments, limiting the effectiveness of school- and community-based prevention programs. Prior research has identified demographic, psychological, and family-related factors associated with service disengagement, but the specific reasons for non-attendance remain underexplored, particularly using data-driven approaches.
The present study, led by García Ramos et al. (2026) addresses this gap by examining reasons for non-attendance among adolescents identified as at risk for suicide within a school-based screening program. By combining traditional psychometric data with Natural Language Processing (NLP) techniques applied to unstructured textual information, the authors aim to provide a more nuanced understanding of disengagement patterns and inform targeted strategies to improve adherence to psychological evaluations.
This study gather data from the EPISAM-School project: a school-based investigation to analyse suicide-related risk factors among adolescents aged 12 to 16 years in Madrid (Spain). Participants were 189 adolescents (72% female) with a mean age of 14.11 years (SD = 1.44) who were identified as at risk for suicide and subsequently failed to attend scheduled psychological assessments. A mixed-data approach was used, including a) psychometric measures assessing emotional symptoms, behavioral problems, family context, and suicide risk indicators; b) free-text explanations documenting reasons for non-attendance, collected during follow-up attempts by clinical or research staff. To identify and classify thematic patterns of non-attendance, a series of NLP techniques were applied, combined with machine learning algorithms. Further statistical analyses were performed to examine associations between non-attendance profiles and psychological, familial, and behavioral variables.
Main reasons to abandon psychological assessment
The NLP-assisted analyses identified several recurrent categories of reasons for non-attendance. These included both parental and adolescent refusal, absence of response, and lack of contact information. Family-related reasons were among the most frequently reported, underscoring the central role of caregivers in facilitating access to mental health care for adolescents.
Moreover, significant associations between specific non-attendance profiles and clinical variables emerged. Adolescents whose non-attendance was linked to avoidance or emotional factors tended to show higher levels of psychological distress and suicide risk indicators. In contrast, logistical reasons were less strongly associated with clinical severity. In addition, the machine learning models demonstrated good classification performance (93%; F1-scores 0.91 – 0.95), supporting the feasibility of using NLP-based approaches to profile disengagement risk.
There is more to this than meets the eye: what could be behind abandoning psychological assessment among high-risk adolescents.
The findings highlight that non-attendance at psychological assessments among high-risk adolescents is a multifactorial phenomenon, shaped by individual, familial, and contextual factors. Importantly, the study shows that not all missed appointments carry the same clinical meaning: some reasons for non-attendance are closely tied to higher vulnerability and may signal increased suicide risk.
The integration of NLP methods with traditional clinical data represents a methodological strength, enabling scalable and systematic analysis of unstructured information that is often underutilized in mental health research. From a clinical perspective, the results suggest that early identification of non-attendance profiles could allow professionals to tailor engagement strategies, such as intensified follow-up, family-focused interventions, or flexible scheduling.
NLP-based techniques may meaningfully enhance the understanding of non-attendance in adolescent suicide prevention programs. By identifying distinct profiles and their clinical correlates, the findings contribute to more precise risk stratification and highlight opportunities for targeted secondary prevention. Incorporating NLP-based tools into school and clinical settings may improve follow-up adherence and ultimately strengthen suicide prevention efforts among adolescents.
