SBIR Phase I: VoxCare: Artificial Intelligence-based Monitoring for Substance Use Indicators in Youth

小型企业创新研究(SBIR)第一阶段项目的更广泛影响是减少每年7400亿美元的物质使用负担。仅在2021年，医疗成本、犯罪和生产力损失只是夺走10万多人生命的问题的一小部分。这项拟议的创新是一项基于人工智能的新型技术，集成到一款移动应用程序中，可以根据语音信号评估药物使用或酒精中毒。物质和酒精使用障碍在我们社会的各个阶层都很普遍。受影响的不仅是用户，受影响的个人也包括美国的每一位公民。防止早期接触非法药物和酒精可以降低在以后的生活中发展成严重成瘾的可能性。通过扰乱正常的大脑发育，各种物质增加了对儿童生活的长期负面影响。鉴于青少年抑郁和焦虑的比例飙升，高效合成物质的广泛可用，以及容易获得毒品，父母们感到没有足够的装备来保护他们的孩子。这个小企业创新研究第一阶段项目将有助于促进几个重要领域的知识，如人工智能能力，利用数字媒体平台进行数据收集，合成音频数据生成，以及使用时间域和频谱域的混合进行音频分析。每个领域都是一个活跃的研究领域，但将所有这些领域结合到一个产品中，用于分析声音来检测各种物质引起的中毒，以前学术界和工业界都没有尝试过。该公司的技术依赖于每种物质使用对神经运动功能的独特影响，这体现在声音产生的生物机械过程中。通过将语音信号与说话者的基线语音进行比较，该技术将标记出通常与醉酒一致的声学特征的异常偏差。语音分析领域集成了一系列特征和信号特征，以从语音信号中提取有价值的见解。主要目标是在手机等无处不在的设备中开发和部署这种基于人工智能的技术，同时确保其在现实生活条件下的性能并实现软实时处理。该奖项反映了NSF的法定使命，并通过使用基金会的智力优势和更广泛的影响审查标准进行评估，被认为值得支持。

The broader impact of this Small Business Innovation Research (SBIR) Phase I project is to reduce the $740B a year burden of substance use. Healthcare costs, crime, and lost productivity are only a fraction of the problem that has claimed over 100,000 lives in 2021 alone. The proposed innovation is a novel AI-based technology integrated into a mobile app that allows the assessment of drug use or alcohol intoxication based on voice signals. Substance and alcohol use disorders are prevalent across every stratum of our society. Not only are the users impacted, but the penumbra of individuals impacted includes every citizen in the US. Preventing early exposure to illicit drugs and alcohol lowers the likelihood of developing severe addiction later in life. By disrupting normal brain development, various substances increase the long-term negative effects on a child’s life. Given the skyrocketing rates of depression and anxiety among teenagers, the wide availability of highly potent synthetic substances, and the ease of access to drugs, parents feel ill-equipped to protect their children.This Small Business Innovation Research Phase 1 project will help advance knowledge in several important areas, such as artificial intelligence capabilities, leveraging digital media platforms for data collection, synthetic audio data generation, and using the mix of temporal and spectral domains for audio analysis. Each area is an active research field, but the combination of all these areas into a product for analyzing voice to detect intoxication caused by various substances has not been previously attempted by academia or industry. The company’s technology relies on the unique impact that each substance use has on neuromotor function, as manifested by the bio-mechanical process of voice production. By comparing the voice signal against the speakers’ baseline voice, the technology will flag abnormal deviations of acoustic features that are typically consistent with intoxication. The field of voice analysis integrates a range of features and signal characteristics to extract valuable insights from voice signals. The primary objectives are to develop and deploy this AI-based technology in ubiquitous devices such as mobile phones while ensuring its performance in real-life conditions and achieving soft real-time processing.This award reflects NSF's statutory mission and has been deemed worthy of support through evaluation using the Foundation's intellectual merit and broader impacts review criteria.