Using Real-time Speech Translation for Medication Therapy Management of Spanish S

描述(申请人提供)：这项SBIR研究的目的是评估使用实时语音翻译设备来促进主要说西班牙语的糖尿病患者进行MTM的可行性。我们计划通过以下方式实现这一目标：(1)使用加州大学旧金山分校药剂师的实际数据和主题专业知识建立药剂师与患者对话的模型，(2)研究如何改进Fluential现有的语音翻译技术，以接受更长的输入-药剂师与患者对话的平均话语长度为15字，而Fluential之前实施的护士与患者对话技术的平均话语长度为5字，以及(3)评估患者回授技术，这是一个药剂师确保患者理解药房访问期间建议的行动计划的过程。这项SBIR研究应用的具体目标是：1.建立一个语音翻译系统，可以接受长达20个单词的话语，而不会降低准确率。2.确定使用该系统的药师和患者的满意度水平。Fluential预计Likert量表的平均分数为4分(满分5分)，将进入第二阶段。3.比较使用Fluential系统进行回教的准确性。Fluential预计90%的短语将被准确翻译到第二阶段。为了建立初始系统，Fluential打算收集25个药剂师-患者对话，使用目前指导这些对话的两名加州大学旧金山分校的药剂师。一旦该系统建立，Fluential将对该系统进行12次药剂师与患者的互动评估。由于时间限制，Fluential受到其可以收集的互动数量的限制；因此，Fluential和UCSF正在努力确定语音翻译系统是否是药剂师与患者互动的有效工具，特别是当它与回教有关时。 公共卫生相关性：在这项申请中，Fluential，Inc.和UCSF建议评估使用实时语音翻译设备来促进以西班牙语为主要语言的有限英语(LEP)糖尿病患者的药物治疗管理(MTM)的可行性。

DESCRIPTION (provided by applicant): The purpose of this SBIR study is to evaluate the feasibility of using a real-time speech translation device to facilitate MTM of patients with diabetes who primarily speak Spanish. We plan to accomplish this by (1) building a model of pharmacist-patient conversations using actual data and subject matter expertise from UCSF pharmacists, (2) researching ways to improve Fluential's existing speech translation technology to accept longer inputs - 15 word average utterances length for pharmacist-patient conversation compared to the 5 word average utterances for nurse-patient conversation technology that Fluential has previously implemented, and (3) evaluating the technology for patient teach-back, a process whereby the pharmacist ensures that the patient understands the action plan recommended during the pharmacy visit. The specific aims in this SBIR research application are to: 1. Build a speech translation system that can accept long (up to 20 words long) utterances without any degradation in accuracy. 2. Determine the level of satisfaction of pharmacists and patients who use the system. Fluential expects an average score of 4 out of 5 on a Likert scale to move forward to Phase II. 3. Compare the accuracy between teach-back using the Fluential system. Fluential expects 90 percent of the phrases to be translated accurately to move forward to Phase II. In order to build the initial system, Fluential intends to collect 25 pharmacist-patient dialogues using the two UCSF pharmacists that currently conduct them. Once the system is built, Fluential will evaluate the system on 12 pharmacist-patient interactions. Due to time constraints, Fluential is limited by the number of interactions that it can collect; and as such Fluential and UCSF are trying to determine whether a speech translation system is an effective tool for pharmacist-patient interaction, especially as it pertains to teach-back. PUBLIC HEALTH RELEVANCE: In this application, Fluential, Inc. and UCSF propose to evaluate the feasibility of using a real-time speech translation device to facilitate Medication Therapy Management (MTM) of Limited English Proficient (LEP) patients with diabetes who primarily speak Spanish.