心理健康应用的长寿性:最大心理健康应用数据库的生存分析
该研究基于MIND数据库对865款心理健康应用进行5年纵向生存分析,发现53.6%的应用被移除,平台独占性(仅安卓)是移除的最主要预测因素,其风险比达2.32。睡眠类应用最长寿,戒烟类应用移除风险最高。
要点
- 仅安卓平台的应用移除风险是双平台应用的2.32倍。
- 睡眠类应用生存期最长,戒烟类应用移除风险最高。
- 5年内53.6%的心理健康应用被从数据库移除。
核心结论
- 心理健康应用市场不稳定,平台独占性(仅安卓)是应用被移除的最主要预测因素。
- 睡眠类应用最长寿,戒烟类应用移除风险最高。
研究设计与样本
- 基于MIND数据库的5年纵向生存分析,纳入865款应用。
- 采用Kaplan-Meier估计和Cox回归评估54个预测因素,并用随机森林分类器识别2年生存特征。
主要发现
- 865款应用中464款(53.6%)在2142天研究期内被移除。
- Cox回归C-index=0.77,随机森林AUC=0.82,平台独占性是最突出特征。
- 仅安卓应用的风险比=2.32,Gini特征重要性=0.12。
关键数值与推荐
- 仅安卓平台应用的风险比=2.32(Cox回归)。
- Cox回归C-index=0.77,随机森林AUC=0.82。
临床怎么用
- 临床选择心理健康应用时,优先考虑同时支持安卓和iOS的双平台应用以降低治疗中断风险。
- 对戒烟类应用需警惕其高移除风险,睡眠类应用相对更稳定。
以上解读依据论文摘要(全文未开放获取或许可不允许翻译转载)。
原文
OBJECTIVE: Despite over 10,000 mental health apps available, app marketplaces are unstable, characterized by high attrition rates and technological obsolescence. This study analyzed factors associated with app survival to help patients and clinicians in selecting more durable mental health apps and inform developers in creating more stable ones. METHODS: Using data from the M-Health Index and Navigation Database (MIND; MindApps.org), the authors performed a 5-year longitudinal survival analysis on 865 apps. App longevity was evaluated using Kaplan-Meier estimates, and Cox regression was used to assess the impact of 54 clinical, technical, and commercial predictors of attrition. Additionally, a random forest classifier was used to identify app characteristics associated with 2-year survival status. RESULTS: Of 865 apps identified, 464 (53.6%) were removed from MindApps.org during the 2,142-day study period. Disparities in longevity were observed across clinical targets; aside from the small cohort of apps for schizophrenia, smoking cessation apps presented the highest risk for attrition. Sleep-related apps demonstrated the greatest longevity. Cox regression (C-index=0.77) and an exploratory random forest analysis (area under the receiver operating characteristic curve=0.82) identified platform exclusivity (not working on both Android and iOS) as the most prominent characteristic associated with removal from MindApps.org, with Android-only apps exhibiting a significantly elevated risk, carrying more than double the hazard of attrition (Gini feature importance=0.12; hazard ratio=2.32). CONCLUSIONS: The findings provide a roadmap for identifying durable mental health apps, thereby minimizing the risk of treatment attrition. The predictors of removal from MindApps.org provide critical lessons for the new era of AI-based tools.
引用
- Herpertz J, Dwyer B, Flathers M, et al. The Longevity of Mental Health Apps: A Survival Analysis of the Largest Mental Health App Database. Am J Psychiatry. 2026. PMID: 42711750. doi:10.1176/appi.ajp.20260155
供执业医师工作参考,不替代临床判断;数值与分级以指南原文为准。