Tissue optics-informed blood hemoglobin assessment for advancing sickle cell care

Blood hemoglobin testing is a common clinical laboratory procedure but typically involves invasive sampling, leading to potential blood loss. While noninvasive technologies are emerging, their accuracy often does not match that of clinical laboratory tests. Optical spectroscopy provides reliable hemoglobin measurements but is limited by bulky components, high costs, and slow data acquisition. Digital color-based diagnostics hold potential for point-of-care blood hemoglobin testing but face challenges in achieving accurate and precise color readings due to variations in devices, light conditions, and image formats. To address these limitations, we present a mobile health algorithm combining hyperspectral learning and machine color detection for reliable, noninvasive blood hemoglobin quantification. The palpebral conjunctiva (inner eyelid) is selected as an optimal peripheral tissue site due to its accessibility, uniform microvasculature, and lack of skin pigmentation. First, hyperspectral learning reconstructs a reflectance spectrum from smartphone RGB values of the inner eyelid, eliminating the need for complex optical instruments. Second, informed by tissue optics, spectroscopic analysis of peripheral tissue establishes a diagnostic color chart, enabling machine-readable quantification of subtle color variations associated with blood hemoglobin. The reported method for noninvasive blood hemoglobin assessment is applied to improve care for sickle cell disease patients, as hemoglobin levels are crucial for understanding the disease’s pathophysiology, including sickle cell anemia. This approach demonstrates accuracy comparable to capillary blood sampling tests and ensures device/light/format-agnostic color detection of the inner eyelid. Importantly, a single-shot digital photograph allows noninvasive, real-time hemoglobin readings, making it ideal for resource-limited and at-home settings.