SBIR Phase I: Development of An Accurate Low-Cost Wearable Ultraviolet Dosimeter For The General Population

这项小型企业创新研究（SBIR）第一阶段项目的更广泛影响是为消费者提供一种科学工具来测量和控制他们暴露于紫外线（UV）辐射的程度，从而降低他们患皮肤癌的风险，同时享受阳光的好处，如维生素D和户外活动。在美国，皮肤癌已成为一个主要的公共卫生问题，估计每年有350万人接受治疗，费用超过80亿美元。据估计，美国维生素D缺乏率超过40%，导致抑郁症，心血管疾病和癌症的风险增加。到目前为止，准确测量紫外线照射仍然局限于研究实验室。该项目旨在实现紫外线剂量测量的科学突破，并为消费者带来实验室级的技术。该项目旨在通过结合多种检测器，机器学习算法和定制校准来实现紫外线剂量测量的科学突破。我们预计，这一突破将导致精度大幅提高，更接近实验室级设备，同时保持我们的仪器的尺寸和成本符合消费者？预期科学上的挑战是让紫外线传感器在测量大多数太阳光谱时保持准确，并且根据位置，天气和一年中的时间，这些光谱是无限的。我们的策略是远离增量改进（例如滤波器优化）或昂贵的开发（例如全面的光谱仪）。相反，我们正在测试一个假设，即机器学习算法和一小部分精心挑选的探测器的结合将使我们能够建立一个小型低成本的仪器，能够识别当地的太阳光谱，并提供正确校准的实时测量。为了实现这一目标，我们建议应用聚类技术来找到太阳紫外线的代表性光谱，并训练几个探测器来识别它们并校正测量。这个项目的执行需要我们为这个项目聚集的不同合作者之间的顶级研发。

The broader impact of this Small Business Innovation Research (SBIR) Phase I project is to equip consumers with a scientific tool to measure and control their exposure to ultraviolet (UV) radiation, thereby mitigating their risk of getting skin cancer while enjoying the benefits of sunlight such as Vitamin D and outdoors activities. In the United States, skin cancer has become a major public health issue with an estimated 3.5M people being treated each year for a cost of over $8B. The rate of Vitamin D deficiency in the US has been estimated to be over 40% leading to increased risk for depression, cardiovascular disease, and cancer. Up until now, accurate measurement of ultraviolet exposure remains confined to research laboratories. This project aims at carrying a scientific breakthrough in UV dosimetry and bring a laboratory-grade technology to consumers.The proposed project aims to achieve a scientific breakthrough in UV dosimetry by combining several detectors, machine learning algorithms, and customized calibration. We anticipate that this breakthrough will lead to a drastic improvement in accuracy to closer match that of laboratory-grade equipment, while keeping the size and cost of our instrument in line with consumers? expectations. The scientific challenge is to have the UV sensor be accurate when it measures most solar spectra, and there is an infinity of these based on location, weather, and time of the year. Our strategy is to stay away from incremental improvements (e.g. filter optimization) or expensive developments (e.g. a full-blown spectrometer). Instead, we are testing the hypothesis that the combination of machine learning algorithms and a small set of carefully chosen detectors will enable us to build a small low-cost instrument able to identify the local solar spectrum and provide correctly calibrated real-time measurements. To achieve this, we propose to apply clustering techniques to find representative spectra of solar UV and train several detectors to recognize them and correct the measurement. The execution of this project requires top-level R&D among diverse collaborators, whom we have gathered for this project.