AI for Neonatal Resuscitation

Fvital Inc.

AI for Neonatal Resuscitation

Neonatal resuscitation support system using AI-powered video analysis in resource-limited settings.

Tokyo

JPN

For-profit, including B-Corp or similar models

Around 2.3 million newborns die every year around the world. Most of these deaths happen in Sub-Saharan Africa and South Asia. Approximately 900,000 of these deaths occur due to failure to establish breathing at birth, highlighting that improving neonatal resuscitation is an urgent global health priority. International efforts have focused on developing standardized resuscitation algorithms, educational programs, and appropriate medical equipment tailored to low-resource settings. Although these efforts have made progress, they face significant limitations, especially in rural areas where neonatal mortality is high. Key challenges include limited dissemination of resuscitation techniques, difficulties in implementing resuscitation in clinical settings, maintaining resuscitation skills, and lack of post-resuscitation assessments. Healthcare providers in remote areas often struggle to receive consistent training or to apply knowledge effectively under pressure. These persistent gaps highlight the need for innovative and scalable solutions. Traditional approaches have built a foundation, but they alone are not enough to reduce neonatal deaths. We strongly believe that integrating technologies such as AI-powered support systems to guide healthcare providers through effectively reflecting resuscitation procedures and improving outcomes in underserved areas provides a path forward.

We have developed a smartphone-based neonatal resuscitation support system that leverages our proprietary AI-powered video analysis technology. This system non-contactly captures vital signs and identifies resuscitation actions, such as healthcare providers’ movements and the use of specific equipment (patent pending). When healthcare providers upload resuscitation videos recorded on their smartphones, the AI analyzes the footage to automatically generate performance evaluations and detailed chronological records. This enables providers to efficiently review and reflect on resuscitation procedures, even without direct supervision from an instructor. In addition to supporting retrospective review, we aim to aggregate field data to develop two key features: 1) Real-time guidance to support healthcare providers during resuscitation in clinical settings. 2) AI-enhanced educational content to be deployed as a learning support platform for training and capacity-building.

A recent study estimates that increasing the coverage of newborn resuscitation to 90% among births in health facilities could save over 93,000 newborn lives each year at minimal additional cost. Moreover, if resuscitation were provided to 90% of all newborns, regardless of birthplace, as many as 192,000 lives could be saved annually. Our system enables providers to efficiently review and reflect on their resuscitation procedures independently, without the need for an instructor. This supports the development of competence and clinical skills necessary to perform appropriate and timely resuscitation—an essential step toward making a substantial impact on newborn survival. Our solution supports both basic and advanced newborn resuscitation, focusing on helping neonates breathe at birth. Third-party research shows that even basic resuscitation (e.g., stimulation, bag-valve-mask ventilation) can prevent up to 30% of newborn deaths. By equipping healthcare providers to deliver timely, appropriate care even at the most basic level, we can dramatically improve newborn survival outcomes, especially in low-resource settings.