A Deep Learning-Based System for the Assessment of Dental Caries Using Colour Dental Photographs.

D1ental caries remains the most common chronic disease in childhood, affecting almost half of all children globally. Dental care and examination of children living in remote and rural areas is an ongoing challenge that has been compounded by COVID. The development of a validated system with the capacity to screen large numbers of children with some degree of automation has the potential to facilitate remote dental screening at low costs. In this study, we aim to develop and validate a deep learning system for the assessment of dental caries using color dental photos. Three state-of-the-art deep learning networks namely VGG16, ResNet-50 and Inception-v3 were adopted in the context. A total of 1020 child dental photos were used to train and validate the system. We achieved an accuracy of 79% with precision and recall respectively 95% and 75% in classifying 'caries' versus 'sound' with inception-v3.

A Mobile Phone Informational Reminder to Improve Eye Care Adherence Among Diabetic Patients in Rural China: A Randomized Controlled Trial.

PURPOSE: To determine whether short message service (SMS) reminders improve adherence to scheduled ocular examinations among patients with diabetes in rural China. DESIGN: Randomized controlled trial. METHODS: This study enrolled consecutive patients with diabetes scheduled for eye examinations at 5 hospitals in low-income areas of Guangdong, China from March 1, 2015 to May 31, 2016. Participants were randomized (1:1) to receive automated SMS reminders containing information about diabetic retinopathy (DR) 1 week and 3 days prior to scheduled eye appointments (Intervention) or to appointments without reminders (Control). Regression models following intention-to-treat principles were used to estimate the association between the main outcome (attendance within ± 1 week of scheduled visit) and membership in the Intervention group, with and without adjustment for other potential predictors of follow-up. Secondary outcomes included change in DR knowledge score (1, worst; 5, best) and endline satisfaction with care (3, worst; 15, best). RESULTS: Among 233 patients, 119 (51.1%) were randomized to Intervention (age 59.7 ± 11.3 years, 52.1% men) and 114 (48.9%) to Control (58.7 ± 9.50 years, 49.1% men). All participants provided data for the main study outcome. Attendance at scheduled appointments for the Intervention group (51/119, [42.9%]) was significantly higher than for Controls (16/114, [14.0%], between-group difference 28.8% [95% confidence interval (CI) 17.9%, 39.8%], P < .001). Factors associated with attendance in multiple regression models included Intervention group membership (Relative Risk [RR] 3.04, 95% CI, 1.73-5.33, P < .001) and baseline DR knowledge (RR 1.47, 95% CI 1.21-1.78, P < .001). Improvement in Satisfaction (mean difference 1.08, 95% CI 0.70-1.46, P < .001) and DR knowledge (mean difference 1.30, 95% CI 0.96-1.63, P < .001) were significantly higher for the Intervention group. Total cost of the intervention was US$5.40/person. CONCLUSION: Low-cost SMS informational reminders significantly improved adherence to, knowledge about, and satisfaction with care. Additional interventions are needed to further improve adherence.