SBIR Phase I: Leveraging smartphone data to improve clinical decisions in concussion care

这项小企业创新研究（SBIR）项目的更广泛影响/商业潜力是开发一种更客观的脑震荡后症状测量方法。每年，全世界有4200万人遭受脑震荡，美国每年的直接医疗费用为13亿美元。脑震荡是一种可以从先进的远程监测工具中获益的临床情况。症状追踪（如头痛、头晕、疲劳等）是临床医生在做出脑震荡诊断和康复的关键决策时最依赖的评估。不幸的是，症状可能会随着一天中的时间、活动、睡眠或其他与脑震荡无关的因素而波动。此外，症状评估也容易受到回忆偏差的影响。这些限制导致不完整和不准确的症状评估，阻碍了临床医生正确管理治疗策略的能力。这个SBIR第一阶段项目建议开发一种软件，可以使用个人的智能手机远程监测脑震荡症状。这种数字表型的概念已用于精神健康障碍，但尚未应用于脑震荡。调查数字表型对心理健康的影响的研究表明，通过缩短治疗时间和制定客观措施，可以改善诊断和治疗。将数字表现型应用于脑震荡症状可以解决类似的问题：1)治疗时间和2)客观性。提出的解决方案使用实时监控从智能手机的传感器收集数据。特征工程和监督机器学习技术应用于传感器数据，以开发预测脑震荡症状的模型。目前的提案将利用数字表型文献中的既定技术，但将评估这一新应用的其他指标和技术。该奖项反映了美国国家科学基金会的法定使命，并通过使用基金会的知识价值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/commercial potential of this Small Business Innovation Research (SBIR) project is to develop a more objective measure of symptoms after a concussion. Each year, 42 million individuals worldwide suffer a concussion, and cost $1.3 billion per year in direct medical costs in the United States. Concussions represent a clinical scenario that can highly benefit from advanced remote monitoring tools. Symptom tracking (i.e. headaches, dizziness, fatigue, etc.) is the most relied upon assessment clinicians use for critical decisions regarding concussion diagnosis and rehabilitation. Unfortunately, symptoms can fluctuate based on the time of day, activity, sleep, or other non-concussion related factors. In addition, symptom evaluations often are also susceptible to recall bias. These limitations lead to incomplete and inaccurate symptom evaluations that hamper a clinicians ability to properly manage treatment strategies. This SBIR Phase I project proposes to develop software to remotely monitor concussion symptoms using an individual’s smartphone. This concept of digital phenotyping has been used for mental health disorders but has not yet been applied to concussions. Studies investigating digital phenotyping for mental health demonstrate improved diagnosis and treatment by reducing time to treatment and developing objective measures. Applying digital phenotyping to concussion symptoms can solve similar issues: 1) time to treatment and 2) objectivity. The proposed solution uses real-time monitoring to collect data from a smartphone’s sensors. Feature engineering and supervised machine learning techniques are applied to the sensor data to develop a model to predict concussion symptoms. The current proposal will leverage established techniques from the digital phenotyping literature but will evaluate other metrics and techniques for this novel application.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.