The authors reply:

The authors raise three relevant topics for future research: validating the use of pediatric Rapid Ultrasound for Shock and Hypotension (p-RUSH) in a variety of settings in low- and middle-income countries (LMICs), extending the use of p-RUSH as a monitoring tool for treatment response, and the use of artificial intelligence (AI) to simplify and improve repeatability (1). We agree that these are all relevant next steps that could help improve the care for children with shock using bedside ultrasound. These steps were beyond the scope of our feasibility/reliability study in a low-resource setting, in which we focused to determine the shock phenotype upon presentation. We will briefly reflect on the suggested options for future research. Validation of our results in different clinical settings and populations is important to evaluate the implementation of p-RUSH and identify potential barriers. As we mentioned in our Discussion section, this should also include its impact on clinical decision-making and any changes in morbidity and mortality. This would be a great follow-up, as we do know that one size does not fit all, and different settings have different (diagnostic) needs. Indeed, bedside ultrasound is also a useful tool to monitor treatment response, and for this purpose, a more simplified version, often referred to as focused cardiac ultrasound, is used to monitor fluid responsiveness (2). In our study, we aimed to determine shock phenotype in children upon presentation using the more extensive p-RUSH, as it was intended to initiate the most appropriate treatment (3). AI is a promising tool to incorporate in medical devices, which could lead to improved, simplified, and reliable use (4). Examples of automated calculations using ultrasound exist, such as in bladder scans and even more complex circulatory ultrasounds, which have also been trialed in LMIC (5). We also explored the use of machine learning in our dataset (data not displayed); however, our sample size is too small. Large datasets are necessary, which takes time and careful consideration of the sample size to be generalizable and do not introduce bias. For future purposes, adding AI tools to assist sonographers in determining inferior vena cava collapsibility and typing shock are indeed exciting developments.