The ability to detect surgical site infections (SSI) is a critical need for healthcare worldwide, but is especially important in low-income countries, where there is limited access to health facilities and trained clinical staff. In this paper, we present a new method of predicting SSI using a thermal image collected with a smart phone. Machine learning algorithms were developed using images collected as part of a clinical study that included 530 women in rural Rwanda who underwent cesarean section surgery. Thermal images were collected approximately 10 days after surgery, in conjunction with an examination by a trained doctor to determine the status of the wound (infected or not). Of the 530 women, 30 were found to have infected wounds. The data were used to develop two Convolutional Neural Net (CNN) models, with special care taken to avoid overfitting and address the problem of class imbalance in binary classification. The first model, a 6-layer naïve CNN model, demonstrated a median accuracy of AUC=0.84 with sensitivity=71% and specificity=87%. The transfer learning CNN model demonstrated a median accuracy of AUC=0.90 with sensitivity =95% and specificity=84%. To our knowledge, this is the first successful demonstration of a machine learning algorithm to predict surgical infection using thermal images alone. Clinical Relevance-This work establishes a promising new method for automated detection of surgical site infection.

Background: As the volume of surgical cases in low- and middle-income countries (LMICs) increases, surgical-site infections (SSIs) are becoming more prevalent with anecdotal evidence of antimicrobial resistance (AMR), despite a paucity of data on resistance patterns. Objectives: As a primary objective, this prospective study aimed to describe the epidemiology of SSIs and the associated AMR among women who delivered by cesarean at a rural Rwandan hospital. As secondary objectives, this study also assessed patient demographics, pre- and post-operative antibiotic use, and SSI treatment. Methods: Women who underwent cesarean deliveries at Kirehe District Hospital between September 23rd, 2019, and March 16th, 2020, were enrolled prospectively. On postoperative day (POD) 11 (+/– 3 days), their wounds were examined. When an SSI was diagnosed, a wound swab was collected and sent to the Rwandan National Reference Laboratory for culturing and antibiotic susceptibility testing. Findings: Nine hundred thirty women were enrolled, of whom 795 (85.5%) returned for the POD 11 clinic visit. 45 (5.7%) of the 795 were diagnosed with SSI and swabs were collected from 44 of these 45 women. From these 44 swabs, 57 potential pathogens were isolated. The most prevalent bacteria were coagulase-negative staphylococci (n = 12/57, 20.3% of all isolates), and Acinetobacter baumannii complex (n = 9/57, 15.2%). 68.4% (n = 39) of isolates were gram negative; 86.7% if excluding coagulase-negative staphylococci. No gram-negative pathogens isolated were susceptible to ampicillin, and the vast majority demonstrated intermediate susceptibility or resistance to ceftriaxone (92.1%) and cefepime (84.6%). Conclusions: Bacterial isolates from SSI swab cultures in rural Rwanda predominantly consisted of gram-negative pathogens and were largely resistant to commonly used antibiotics. This raises concerns about the effectiveness of antibiotics currently used for surgical prophylaxis and treatment and may guide the appropriate selection of treatment of SSIs in rural Rwanda and comparable settings.

Background: In many developing countries, rheumatic heart disease (RHD) is diagnosed at an advanced stage and requires surgery for patient survival. However, access to cardiac surgery in this context is limited and often provided through partnerships, requiring centralized patient data systems for monitoring and follow-up. Objectives: This study used data from a national postoperative RHD registry to analyze clinical outcomes of Rwandan patients who received surgery between 2006 and 2017. Methods: The RHD registry was created in 2017 using data compiled from Rwanda Ministry of Health and RHD surgery partners. We extracted pre- and post-operative data on patients who were alive and in care. We excluded patients who died or were lost to follow-up, as their data was not collected in the registry. We evaluated the association between demographic, surgical, and follow-up characteristics and most recent patient symptoms, categorized by New York Heart Association (NYHA) class. Findings: Among the 191 patients eligible for inclusion in this study, 107(56.0%) were female, 110(57.6%) were adults at the time of surgery (>15 years), and 128(67.4%) had surgery in Rwanda. Most patients (n = 166, 86.9%) were on penicillin prophylaxis. Of the patients with mechanical valves, 47(29.9%) had therapeutic International Normalized Ratio values. 90% of patients were asymptomatic (NYHA I) at the time of most recent visit. NYHA class was not significantly associated with any of the considered variables. The median length of follow-up for patients was four years (IQR: 2, 5 years). Conclusion: This study shows both the feasibility and challenges of creating a RHD registry 11 years after the national initiation of RHD surgeries. Most patients captured in the registry are asymptomatic; however, collecting details on patients who had died or were lost to follow-up has proven difficult. Implementing strategies to maintain a complete and up-to-date registry will facilitate follow-up for pre- and postoperative patients.

Introduction Cervical cancer is among the most common cancers and is the fourth most common cause of cancer death in women worldwide [1]. Women in low-and middle-income countries (LMICs) disproportionately bear the burden of cervical cancer; 85% of cervical cancer morbidity and 88% of cervical cancer mortality occur in this region [2-4]. In East Africa, among all types of cancers in women, cervical cancer is the leading cause of morbidity and mortality with 52,633 new cases and 37,017 deaths estimated in 2018 [5]. Without adequate investment in cervical cancer control, these rates are only expected to rise [2]. Treatment for cervical cancer is critical for control and secondary disease prevention in LMICs [2]. However, most LMICs have limited infrastructure and human resource capacity to support surgical screening and subsequent treatment with radiotherapy, evidenced by the lack of trained health personnel and inadequate availability of treatment equipment [2]. Where services are available, the cost of treatment often prohibits access [6, 7]. Further, issues such as late presentation at diagnosis, low pretreatment performance status, which indicates a patient's ability to tolerate chemotherapy, lack of adherence to treatment or post-treatment follow-up, and low quality of care worsen patient outcomes [2, 8-11]. Among important programmatic and patient-related aspects of cervical cancer treatment is post-treatment follow-up. Women receiving therapy for invasive cancer

Introduction: Antimicrobial resistance (AMR) is a global public health threat. Worse still, there is a paucity of data from low- and middle-income countries to inform rational antibiotic use. Objective: Assess the feasibility of setting up microbiology capacity for AMR testing and estimate the cost of setting up microbiology testing capacity at rural district hospitals in Rwanda. Methods: Laboratory needs assessments were conducted, and based on identified equipment gaps, appropriate requisitions were processed. Laboratory technicians were trained on microbiology testing processes and open wound samples were collected and cultured at the district hospital (DH) laboratories before being transported to the National Reference Laboratory (NRL) for bacterial identification and antibiotic susceptibility testing. Quality control (QC) assessments were performed at the DHs and NRL. We then estimated the cost of three scenarios for implementing a decentralized microbiology diagnostic testing system. Results: There was an eight-month delay from the completion of the laboratory needs assessments to the initiation of sample collection due to the regional unavailability of appropriate supplies and equipment. When comparing study samples processed by study laboratory technicians and QC samples processed by other laboratory staff, there was 85.0% test result concordance for samples testing at the DHs and 90.0% concordance at the NRL. The cost for essential equipment and supplies for the three DHs was $245,871. The estimated costs for processing 600 samples ranged from $29,500 to $92,590. Conclusion: There are major gaps in equipment and supply availability needed to conduct basic microbiology assays at rural DHs. Despite these challenges, we demonstrated that it is feasible to establish microbiological testing capacity in Rwandan DHs. Building microbiological testing capacity is essential for improving clinical care, informing rational antibiotics use, and ultimately, contributing to the establishment of robust national antimicrobial stewardship programs in rural Rwanda and comparable settings.