METHODS: An app was developed to collect images and patient reported outcome measures (PROMS) including Pain score and the Hand Disability in Systemic Sclerosis-Digital Ulcers (HDISS-DU) questionnaire. Participants photographed their lesion(s) each day for 30 days and uploaded images to a secure repository. Lesions were analysed both manually and automatically, using a machine learning approach.
RESULTS: 25 patients with SSc-related digital lesions consented of whom 19 completed the 30-day study, with evaluable data from 27 lesions. Mean (standard deviation [SD]) baseline Pain score was 5.7 (2.4) and HDISS-DU 2.2 (0.9), indicating high lesion and disease-related morbidity. 506 images were used in the analysis (mean number of used images per lesion 18.7, SD 8.3). Mean (SD) manual and automated lesion areas at day 1 were 11.6 (16.0) and 13.9 (16.7) mm2 respectively. Manual area decreased by 0.08mm2 per day (2.4mm2 over 30 days) and automated area by 0.1mm2 (3.0mm2 over 30 days). Average gradients of manual and automated measurements over 30 days correlated strongly (r = 0.81). Manual measurements were on average 40% lower than automated, with wide limits of agreement.
CONCLUSIONS: Even patients with significant hand disability were able to use the app. Automated measurement of finger lesions could be valuable as an outcome measure in clinical trials.
方法:开发了一个应用程序来收集图像和患者报告的结果指标(PROMS),包括系统性硬化数字溃疡(HDISS-DU)问卷中的疼痛评分和手部残疾。参与者每天拍摄他们的病变,持续30天,并将图像上传到安全存储库。手动和自动分析病变,使用机器学习方法。
结果:25例SSc相关数字病变患者同意,其中19例完成了30天的研究,来自27个病变的可评估数据。平均(标准差[SD])基线疼痛评分为5.7(2.4)和HDISS-DU2.2(0.9),表明高病变和疾病相关发病率。在分析中使用了506张图像(每个病变的平均使用图像数18.7,SD8.3)。在第1天的平均(SD)手动和自动损伤面积分别为11.6(16.0)和13.9(16.7)mm2。手动面积每天减少0.08mm2(30天内减少2.4mm2),自动面积0.1mm2(30天内减少3.0mm2)。30天内手动和自动测量的平均梯度高度相关(r=0.81)。手动测量平均比自动测量低40%。有广泛的协议限制。
结论:即使有严重手部残疾的患者也能够使用该应用程序。手指病变的自动测量作为临床试验中的结果测量可能是有价值的。