Contextualized daily prediction of lapse risk in opioid use disorder by digital phenotyping

项目总结 阿片类药物使用障碍日益普遍，给患者及其家人带来毁灭性的后果和代价 家人、朋友和社区。阿片类药物和其他物质使用障碍(SUD)的现有治疗方法 成功地保持清醒。绝大多数SUD患者在一年内复发。关键的是，他们经常未能做到 发现他们复发风险的每天动态变化，并且没有充分利用他们培养的技能或利用通过持续护理获得的支持。该项目的广泛目标是开发和提供一个高度情景化的失误风险预测模型，用于预测寻求戒毒的人每天发生阿片类药物和其他药物使用失误的可能性。这个失误风险预测模型将在成瘾-全面健康增强支持系统(A-CHESS)移动应用程序中提供，该应用程序已由RCT建立，作为一种最先进的mHealth系统，用于为酒精和物质使用障碍提供持续护理服务。 为了实现这些广泛的目标，由480名患有阿片类药物使用障碍的参与者组成的多样化样本 禁欲将被招募。这些参与者将接受为期12个月的康复跟踪，并进行观察 最早发生在戒断后一周，最晚发生在戒断后18个月 样本。建立良好的远端静态复发风险信号(例如，成瘾严重程度、并存的精神病理)将是 根据摄入量进行测量。一系列更接近的、随时间变化的阿片类药物(和其他药物使用)失误风险信号也将 通过参与者的智能手机收集的。这些信号包括每两个月一次的自我报告调查，每日生态 即时评估、每日视频恢复“签到”、语音电话和短信记录、短信 内容、时刻位置(通过智能手机GPS和定位服务)、体育活动(通过智能手机 传感器)，以及移动A-Chess恢复支持应用程序的使用。这些风险信号的预测力将是 通过将它们固定在已知人员、地点、日期和时间的人际环境中进一步增强 支持或贬低参与者的禁欲努力。将使用机器学习方法来训练、验证和测试基于这些背景静态和动态风险信号的阿片类药物(和其他药物)失误风险预测模型。 这些失误风险预测模型将提供参与者特定的、逐日的阿片类药物戒断者使用阿片类药物(或其他药物)的概率预测。这些失误风险预测模型将在项目完成后正式加入A-CHESS持续护理移动应用程序，用于临床护理。这些项目目标使A-CHESS能够为患者提供持续的复发预防和康复支持、信息和风险监测。与传统的持续护理相比，A-CHESS将提供个性化的护理，并在最需要的时刻可用和实施。综合实时风险预测有很大希望通过适应性地使用这些持续护理服务来鼓励持续恢复。

PROJECT SUMMARY Opioid use disorder is increasingly widespread, leading to devastating consequences and costs for patients and their families, friends, and communities. Available treatments for opioid and other substance use disorders (SUD) are not successful at sustaining sobriety. The vast majority of people with SUD relapse within a year. Critically, they often fail to detect dynamic, day-by-day changes in their risk for relapse and do not adequately employ skills they developed or take advantage of support available through continuing care. The broad goals of this project are to develop and deliver a highly contextualized, lapse risk prediction models for forecasting day-by-day probability of opioid and other drug use lapse among people pursuing drug abstinence. This lapse risk prediction model will be delivered within the Addiction-Comprehensive Health Enhancement Support System (A-CHESS) mobile app, which has been established by RCT as a state-of-the-art mHealth system for providing continuing care services for alcohol and substance use disorders. To accomplish these broad goals, a diverse sample of 480 participants with opioid use disorder who are pursing abstinence will be recruited. These participants will be followed for 12 months of their recovery, with observations occurring as early as one week post-abstinence and as late as 18 months post-abstinence across participants in the sample. Well-established distal, static relapse risk signals (e.g., addiction severity, comorbid psychopathology) will be measured on intake. A range of more proximal, time-varying opioid (and other drug use) lapse risk signals will also be collected via participants’ smartphones. These signals include self-report surveys every two months, daily ecological momentary assessments, daily video recovery “check-ins”, voice phone call and text message logs, text message content, moment-by-moment location (via smartphone GPS and location services), physical activity (via smartphone sensors), and usage of the mobile A-CHESS Recovery Support app. The predictive power of these risk signals will be further increased by anchoring them within an inter-personal context of known people, locations, dates, and times that support or detract from participants’ abstinence efforts. Machine learning methods will be used to train, validate, and test opioid (and other drug) lapse risk prediction models based on these contextualized static and dynamic risk signals. These lapse risk prediction models will provide participant specific, day-by-day probabilistic forecast of a lapse to opioid (or other drug) use among opioid abstinent individuals. These lapse risk prediction models will be formally added to the A-CHESS continuing care mobile app at the completion of the project for use in clinical care. These project goals position A-CHESS to make relapse prevention and recovery support, information, and risk monitoring available to patients continuously. Compared to conventional continuing care, A-CHESS will provide personalized care and be available and implemented during moments of greatest need. Integrated real-time risk prediction holds substantial promise to encourage sustained recovery through adaptive use of these continuing care services.