npj Digital Medicine volume 5, Article number: 146 (2022 ) Cite this article
Hypoxemia, a medical condition that occurs when the blood is not carrying enough oxygen to adequately supply the tissues, is a leading indicator for dangerous complications of respiratory diseases like asthma, COPD, and COVID-19. While purpose-built pulse oximeters can provide accurate blood-oxygen saturation (SpO2) readings that allow for diagnosis of hypoxemia, enabling this capability in unmodified smartphone cameras via a software update could give more people access to important information about their health. Towards this goal, we performed the first clinical development validation on a smartphone camera-based SpO2 sensing system using a varied fraction of inspired oxygen (FiO2) protocol, creating a clinically relevant validation dataset for solely smartphone-based contact PPG methods on a wider range of SpO2 values (70–100%) than prior studies (85–100%). We built a deep learning model using this data to demonstrate an overall MAE = 5.00% SpO2 while identifying positive cases of low SpO2 < 90% with 81% sensitivity and 79% specificity. We also provide the data in open-source format, so that others may build on this work.
Smartphone-based SpO2 monitors, especially those that rely only on built-in hardware with no modifications, present an opportunity to detect and monitor respiratory conditions in contexts where pulse oximeters are less available. Smartphone-based solutions for monitoring blood oxygen saturation have been explored previously, employing various solutions used to gather and stabilize the PPG signal1, augment the IR-filtered broad-band camera sensor2, and filter the resultant signal for noise or outlier correction3. Some solutions require extra hardware, such as a color filter or external light source1,2,4,5,6, whereas others rely only on the in-built smartphone hardware and employ software techniques to process the PPG signal3,7,8,9,10,11. These prior works indicate that there is potential for smartphone-based SpO2 monitors to fill gaps in access to care, but lack validation data on a full range of clinically relevant SpO2 levels. Prior evaluation techniques for these smartphone-based studies have been limited to a minimum of 80% SpO2 using techniques such as breath-holding, which is limited to short durations of data collection due to participant discomfort, limiting the clinical applicability of the findings. The US Food and Drug Administration (FDA) recommends cleared reflectance pulse oximeter devices achieve <3.5% error across the full range of clinically relevant data of 70–100%12,13. To our knowledge, our study is the first to evaluate unmodified smartphone-based pulse oximetry on this range of SpO2 data using a Varied Fractional Inspired Oxygen (Varied FiO2) study procedure, as shown in (Fig. 1).