Protocol for a cluster randomised controlled trial to evaluate the effectiveness of digital health interventions in improving non-communicable disease management during the pandemic in rural Pakistan (preprint)

Background: The COVID-19 pandemic has revealed gaps in global health systems, especially in the low- and middle-income countries (LMICs). Evidence shows that patients with non-communicable diseases (NCDs) are at higher risk of contracting COVID-19 and suffering direct and indirect health consequences. Considering the future challenges such as environmental disasters and pandemics to the LMICs health systems, digital health interventions (DHI) are well poised to strengthen health care resilience. This study aims to implement and evaluate a comprehensive package of DHIs of integrated COVID-NCD care to manage NCDs in primary care facilities in rural Pakistan. Methods The study is designed as a pragmatic, parallel two-arm, multi-centre, cluster randomised controlled trial. We will randomise 30 primary care facilities in three districts of Punjab, where basic hypertension and diabetes diagnosis and treatment is provided, with a ratio of 1:1 between intervention and control. In each facility, we will recruit 50 patients who have uncontrolled hypertension. The intervention arm will receive training on an integrated COVID-NCD guideline, and will use a smartphone app-based telemedicine platform where patients can communicate with health providers and peer-supporters, along with a remote training and supervision system. Usual care will be provided in the control arm. Patients will be followed up for 10 months. Our primary indicator is systolic blood pressure measured at 10 months. A process evaluation guided by implementation science frameworks will be conducted to explore implementation questions. A cost-effectiveness evaluation will be conducted to inform future scale up in Pakistan and other LMICs. Discussion: Our study is one of the first randomised controlled trials to evaluate the effectiveness of DHIs to manage NCDs to strengthen health system resilience in LMICs. We will also evaluate the implementation process and cost-effectiveness to inform future scale-up in similar resource constrained settings.

Contrastive Attention Network with Dense Field Estimation for Face Completion (preprint)

Most modern face completion approaches adopt an autoencoder or its variants to restore missing regions in face images. Encoders are often utilized to learn powerful representations that play an important role in meeting the challenges of sophisticated learning tasks. Specifically, various kinds of masks are often presented in face images in the wild, forming complex patterns, especially in this hard period of COVID-19. It's difficult for encoders to capture such powerful representations under this complex situation. To address this challenge, we propose a self-supervised Siamese inference network to improve the generalization and robustness of encoders. It can encode contextual semantics from full-resolution images and obtain more discriminative representations. To deal with geometric variations of face images, a dense correspondence field is integrated into the network. We further propose a multi-scale decoder with a novel dual attention fusion module (DAF), which can combine the restored and known regions in an adaptive manner. This multi-scale architecture is beneficial for the decoder to utilize discriminative representations learned from encoders into images. Extensive experiments clearly demonstrate that the proposed approach not only achieves more appealing results compared with state-of-the-art methods but also improves the performance of masked face recognition dramatically.