基于智能手机图像的非侵入性即时镰状细胞病筛查。

Non - Invasive, smartphone image-based screening for sickle cell disease at the point-of-need.

作者信息

Vital Eudorah F, LiCalzi Meredith Haak, Mannino Robert G, McGann Patrick T, Lam Wilbur A

机构信息

Wallace H. Coulter Department of Biomedical Engineering, Georgia Institute of Technology and Emory University, Atlanta, Georgia, 30332, US.

Sanguina, Inc, Atlanta, Georgia, 30092, US.

出版信息

Heliyon. 2025 Jan 10;11(2):e41830. doi: 10.1016/j.heliyon.2025.e41830. eCollection 2025 Jan 30.

DOI:10.1016/j.heliyon.2025.e41830

PMID:39897844

原文链接:https://pmc.ncbi.nlm.nih.gov/articles/PMC11787632/

Abstract

Leveraging the increasing accessibility of smartphones in healthcare settings, we developed a smartphone app aimed at enhancing sickle cell disease (SCD) screening, particularly in resource-limited settings. Our application provides accurate and non-invasive SCD screening with instant results at the point-of-need. The app operates by analyzing patient fingernail images via a smartphone image to gauge anemia severity and by using targeted inquiries to identify SCD-related symptoms. These inputs collectively generate an CD age and atient profile-based ikelihood stimation (SIMPLE) score, which estimates disease probability. The accuracy of the score depends on two inputs: the patient's anemia status and their health survey responses. We tested the app on 485 pediatric patients at Children's Healthcare of Atlanta (CHOA), achieving an overall sensitivity of 74 % and specificity of 76 % in screening for SCD among children aged 6 months to 21 years. Notably, the app demonstrated enhanced performance in the target demographic, with 100 % sensitivity and 75 % specificity for screening SCD in children aged 6 months to 5 years. This cost-effective and scalable app efficiently pinpoints and stratifies individuals, particularly those who missed early screening, for formal screening programs.

摘要

利用智能手机在医疗环境中日益普及的优势，我们开发了一款智能手机应用程序，旨在加强镰状细胞病（SCD）筛查，尤其是在资源有限的环境中。我们的应用程序提供准确且无创的SCD筛查，并在需要时即时给出结果。该应用程序通过智能手机图像分析患者指甲图像以评估贫血严重程度，并通过针对性询问来识别与SCD相关的症状。这些输入共同生成一个基于患者年龄和病历的疾病可能性估计（SIMPLE）分数，用于估计患病概率。该分数的准确性取决于两个输入：患者的贫血状况及其健康调查回复。我们在亚特兰大儿童医疗保健中心（CHOA）对485名儿科患者进行了该应用程序的测试，在筛查6个月至21岁儿童的SCD时，总体灵敏度达到74%，特异性达到76%。值得注意的是，该应用程序在目标人群中表现更佳，在筛查6个月至5岁儿童的SCD时，灵敏度为100%，特异性为75%。这款经济高效且可扩展的应用程序能有效地确定个体并进行分层，特别是那些错过早期筛查的个体，以便他们参加正式的筛查项目。