BackgroundIn recent years， the prevalence of depression among adolescents has risen steadily， alongside an increasing prominence of non-suicidal self-injury （NSSI） behaviors in this population， which may elevate suicide risk and constitute a serious public health problem. Negative emotions such as anxiety and depression are closely related to NSSI behaviors. Previous studies have predominantly relied on retrospective reports， limiting the ability to dynamically capture temporal relationships between emotional fluctuations and NSSI behaviors. Moreover， cognitive behavioral therapy （CBT） often lacks targeted design to address proximal triggers in interventions for NSSI behaviors. ObjectiveTo verify that anxiety as a proximal trigger factor for NSSI behaviors in adolescent patients with depression， and to evaluate the efficacy of modified CBT integrated with ecological momentary assessment （EMA） data in alleviating their anxiety and self-injury urges. MethodsA prospective cohort study design was adopted. A total of 132 adolescent patients with a history of NSSI behaviors who were treated at Jiujiang Fifth People's Hospital from January to December 2024 and met the diagnostic criteria for depression in the Diagnosed and Statistical Manual of Mental Disorders， fifth edition （DSM-5）. A smartphone-based EMA application to conduct natural situation emotions on participants for 14 consecutive days. The monitoring period spanned from 10∶00 to 22∶00 daily， with randomized assessments pushed every 2 hours. The assessment tools included the Self-rating Anxiety Scale （SAS） and the negative affect subscale of the Positive and Negative Affect Schedule （PANAS）， with real-time recording of NSSI behaviors and emotional states before and after their occurrence. Repeated measures analysis of variance was used to compare the dynamic changes in the scores of each scale before and after the episodes of NSSI behaviors. A modified CBT was administered to 83 participants who completed EMA data collection， with sessions conducted once weekly for 12 weeks. Anxiety levels were accessed using the SAS， and the self-injury urges was evaluated using the Ottawa Self-injury Inventory （OSI） before and after the intervention. ResultsEMA data revealed that SAS scores were significantly higher during the 1-2 hours before NSSI episodes compared to baseline periods ［（56.19±11.06）vs.（52.83±10.25），P<0.01］. SAS scores were positively correlated with the NSSI behavioral scores （r=0.460，P<0.01，95% CI：0.310-0.580）. After receiving modified CBT intervention， adolescent patients with depression demonstrated statistically significant decreases in both SAS scores ［（52.30±8.10） vs.（48.70±7.30），t（82）=4.820，P<0.01，Cohen's d=0.420］ and the OSI self-injury impulse subcale scores ［（12.80±2.70） vs.（9.60±2.50），t（82）=5.170，P<0.01，Cohen's d=0.510］ compared to their pre-intervention levels. ConclusionAnxiety may serve as a proximal trigger for NSSI behaviors in adolescent patients with depression. Modified CBT integrating EMA data could potentially alleviate their anxiety level and self-injury urges. ［Funded by Science and Technology Plan Project of Jiangxi Provincial Health Commission （number， SKJP220227629）］