Smartphone-Based Blood Oxygen Saturation Measurement for Respiratory Distress Evaluation

Clinical Research Results Abstract Research Idea Abstract BackgroundPulse oximeters, which provide a noninvasive measure of blood-oxygen saturation (SpO2), are becoming more ubiquitous but are still not available in many settings, motivating the need for more accessible means of hypoxia screening. We have developed an algorithm that can detect hypoxia events (below 90% SpO2) with 80% sensitivity and specificity using a photoplethysmogram (PPG) signal gathered by an unmodified smartphone camera [Hoffman].Outline of the research questionIn order to expand applicability of this technology to clinical and home monitoring scenarios, 3 broad axes of validation need to be addressed: accuracy, cross-compatibility, and usability [Mariakakis]. In this research project, we propose to conduct an investigation of the accuracy of the tool in the clinical setting, comparing our software solution against an FDA-cleared clinical oximeter across patients with a broad range of oxygen saturations, ages, underlying conditions, and skin tones.Possible MethodologyWe plan to enroll patients presenting to the emergency department or admitted to the hospital with respiratory distress (chronic or acute). After informed consent, SpO2 will be measured with both the gold standard pulse oximeter and the phone application. The following data will be recorded: SpO2 with both devices, supplemental oxygen status, age, gender, race/ethnicity, diagnosis, and Fitzpatrick-scale skin tone. Signal quality and predicted SpO2 level will be assessed, compared against the reference oximeter using Bland-Altman plots, Pearson’s correlation coefficient, and ROC curves for hypoxia classification of the phone-based measure (SpO2 <90%). Pre-specified stratified analyses will evaluate the accuracy in different age categories and skin tones. Future work will focus on acceptability, usability and implementation, particularly in low-resource settings.Questions to DiscussAre there any other measures we should plan to gather relevant to SpO2?Are there study changes to further improve the validity of the approach?What study design considerations to increase feasibility in a busy emergency department? Service Development & Evaluation Abstract Declaration of Interest Jason Hoffman and Shwetak Patel are inventors of US patent application 17/164,745 covering systems and methods for SpO2 classification using smartphones. References and Clinical Trial Registry Information Hoffman, J. S., Viswanath, V., Ding, X., Thompson, M. J., Larson, E. C., Patel, S. N., & Wang, E. (2021). Smartphone Camera Oximetry in an Induced Hypoxemia Study. ArXiv:2104.00038 [Cs]. http://arxiv.org/abs/2104.00038Mariakakis, A., Wang, E., Patel, S., & Goel, M. (2019). Challenges in Realizing Smartphone-Based Health Sensing. IEEE Pervasive Computing, 18(2), 76–84. https://doi.org/10.1109/MPRV.2019.2907007