Remote Monitoring and Detecting of Tardive Dyskinesia for Improving Patient Outcomes

远程监测和检测迟发性运动障碍以改善患者预后 迟发性运动障碍(TDD)是抗精神病药物使用的一种常见的衰弱副作用。最显著的特点是 不自觉的面部动作，如做鬼脸，不自觉的嘴唇，嘴巴和舌头的动作，以及眨眼， TDD很难治疗，而且有可能不可逆转。精神病学家和其他精神健康专业人士敏锐地 意识到患有TDD的患者所经历的损害和残疾。TDD的早期发现是 至关重要，以便能够采取适当的干预措施。实施的干预措施是密切相关的 与了解病人的服药依从性有关。即使是最有资格的诊断专家也很难 每年投入20-25分钟的面对面时间，频率为4至6次，以提供 病人：1)“积极监测”，2)讨论结果，3)预期的药物变化和使用说明 随着当今对每一位精神健康专业人员的迫切需求。这是越来越具有挑战性的 由于大流行，远程医疗和患者人数增加，人力资源减少。 不幸的是，尽管专业人士尽了最大努力，但在这个过程中往往为时已晚，而且行动不由自主 是永久性的。目前，有20万人在服用抗TDD药物，价格分别为6万美元和1.05万美元 每年，这一数字每年都在迅速增加。一种TDD自动检测和准确跟踪的方法 将能够及时干预，避免患者耻辱、降低生活质量和昂贵的持续治疗 用于永久性TDD。 2020年抗精神病药物处方超过5000万张，报告的TDD患病率在13%至 24%。风险随着年龄的增长、非标签使用和长期接触抗精神病药物而增加。因此， 预防和早期发现是管理TDD的关键。然而，目前监测患者的方法 需要在不经常亲自探视的情况下观察患者，或由警觉但培训不足的患者自我报告 以及他们的家人。因此，自动化远程遵从性监控和 TDD检测系统。商业化计划中提出了我们的入市战略。 这个第二阶段的项目计划利用现有的心灵感应和视频访谈数据收集技术 在第一阶段，在对患有拓展性疾病的个人进行分类时，表现出高达77%的歧视，而在3- 由训练有素的临床专业人员组成的专家小组评估相同的视频材料。基于对 第一阶段的数据，我们在这里建议扩大收集和分析额外的300个录像目标 以及对服用抗精神病药物的个人进行的5分钟视频采访。一半的采访将与 确诊为TDD的患者和未诊断为TDD的患者。这项研究的参与者将 被招募以确保男女平等分配以及种族和种族方面的 具有代表性的样本。 拟议的数据收集战略将提供敲定和部署强大的 有监督的机器学习派生视频和音频分析工具来检测TDD。检测工具将是 使用80%的收集视频数据作为训练集创建，并在剩余20%的保留为 控制装置。基于其他受监督机器学习训练集的行业经验和 要收集的数据量，我们设定了一个目标，即识别TDD阳性和TDD阴性的成功率为90% 控制组中的参与者。 一旦检测工具完成，该项目将通过将对该工具的访问整合到现有的 另一款是智能手机应用iRxRMinder，用于数据收集和监测服药依从性 临床干预所需的关键组件。IRxRMinder平台将患者直接连接到 研究人员和他们的电子记录。修改后的APP将在实验室进行测试，以确保界面 使用方便。然后，该第二阶段项目将使用iRxRminder平台来支持自我 慢性阻塞性肺病患者服药的管理和TDD及其他症状监测 精神疾病。在第一阶段确立可行性后，我们建议进行为期六个月的临床试验， 将对参与者进行监测，以便及早发现TDD(并确认没有TDD，从而避免 不必要的诊断时间)以及2)高依从性的目标，3)改进症状控制和 副作用，以及4)医疗团队更具侵略性和更频繁的治疗反应。统计检验 患者和护理团队的易用性将被进行。对收入、治疗轨迹的影响 (检测到的副作用和所做的药物改变的数量)将被评估。算法的成功 在6个月的监测结束时，将检测TDD与人工评估进行比较，这将是 这项技术。

Abstract - Remote Monitoring and Detecting of Tardive Dyskinesia for Improving Patient Outcomes Tardive dyskinesia (TDD) is a common debilitating side effect of antipsychotic use. Characterized most notably by involuntary facial movements such as grimacing, involuntary lip, mouth, and tongue movements, and eye blinking, TDD is difficult to treat and potentially irreversible. Psychiatrists and other mental health professionals are acutely aware of the impairment and disability experienced by patients who develop TDD. Early detection of TDD is critical so that appropriate interventions can be instituted. What interventions are implemented is intimately tied to knowing the patient’s medication adherence. It is difficult for the most qualified diagnosticians to devote the 20-25 minutes of in-person time at the 4 to 6 times per year frequency necessary to provide every patient the 1) “active monitoring,” 2) discussion of results, 3) changes to medication and instructions expected with the urgent demands on every mental health professional today. This is increasingly challenging with the increase in telemedicine and patient populations and decreasing human resources due to the pandemic. Unfortunately, despite professionals’ best efforts, it is often too late in the process and the involuntary movements are permanent. Currently, there are 200,000 individuals taking anti-TDD medications costing $60K and $105K annually and this is increasing rapidly each year. A method for automatic TDD detection and accurate adherence would enable timely intervention and avoid patient stigma, lower quality of life, and expensive ongoing treatment for permanent TDD. Antipsychotic prescriptions exceeded 50 million in 2020 and the reported prevalence of TDD is between 13% and 24%. Risk grows with advancing age, off-label uses, and chronic exposure to antipsychotics. Therefore, prevention and early detection are key to managing TDD. However, current methods for monitoring patients require observation of patients at infrequent in-person visits or self-reporting by vigilant but undertrained patients and their families. Therefore, strong market potential exists for an automated remote adherence monitoring and TDD detection system. Our go-to-market strategy is presented in the commercialization plan. This Phase II project proposes to leverage existing telepsychiatry and video interview data gathering technologies that in Phase I demonstrated up to 77% discrimination in categorizing individuals with TDD compared to a 3- person panel of trained clinical professionals evaluating the same video materials. Based on a power analysis of the Phase I data, we propose here to extend collection and analysis of an additional 300 video recorded AIMS and 5-minute video interviews with individuals taking anti-psychotic medications. Half of the interviews will be with individuals living with diagnosed TDD and the other without a diagnosis of TDD. The participants in the study will be recruited to ensure an equal distribution of females and males as well as an ethnically and racially representative sample. The proposed data gathering strategy will provide the source material necessary to finalize and deploy a powerful supervised machine learning derived video and audio analysis tool to detect TDD. The detection tool will be created using 80% of the collected video data as a training set and validated on the remaining 20% reserved as the control set. Based on industry experience with other supervised machine learning training sets and the amount of data to be collected, we set a goal of a 90% success rate in identifying TDD positive and TDD negative participants in the control set. Once the detection tool is complete the project will conclude by incorporating access to the tool into an existing smartphone app, iRxReminder, that is used for data gathering and monitoring of medication adherence, the other critical component required for clinical intervention. The iRxReminder platform links patients directly to researchers and their electronic records. The modified app will be tested in the laboratory to ensure the interface can be easily used. This Phase II project will then use the iRxReminder platform for use in supporting the self- management and TDD and other symptoms monitoring of medication taking by individuals living with chronic mental illnesses. With feasibility established in Phase I, we propose a six-month long clinical trial where participants will 1) be monitored for early detection of TDD (and confirmation of not having TDD, thus avoiding unnecessary diagnostician time) along with 2) goals for high adherence, 3) improved control of symptoms and side effects, and 4) more aggressive and frequent treatment responses by the healthcare team. Statistical tests of the ease-of-use by patients and the care team will be conducted. The impact on revenue, treatment trajectory (number of side effects detected and medication changes made) will be assessed. The success of the algorithm to detect TDD compared to a human assessment at the end of 6-months of monitoring will be a final field test of the technology.