Quick Buzz Feed
BU computer scientist and pediatrician researching potential of AI-powered assistance for ambulance crews
The article opens with a vivid account of a simulated emergency involving a baby experiencing an apparent seizure. Paramedics Khanzada-Marie Kurbanova and Rachel Miller respond to a call from the 'aunt' (played by pediatrician Tehnaz Boyle). They administer care on a gurney inside an ambulance, including fitting a breathing tube and preparing an IV for dextrose, while Kurbanova consults a Boston Medical Center doctor via iPad video call. The simulation emphasizes the real-time decision-making process, the stress of the environment, and the critical interventions needed. This hands-on scenario, involving a high-fidelity mannequin whose vitals are controlled by a SimPad, serves as a crucial data collection method. Boyle is conducting over 500 such observations across various EMS agencies to understand the dynamics and efficacy of emergency pediatric care in the field. The immediate goal is to evaluate if remote physician consultation improves field care, while the long-term vision involves training AI models.
At the core of this extensive research, a significant collaborative effort between Tehnaz Boyle, a pediatric emergency physician at Boston Medical Center, and Deepti Ghadiyaram, a BU College of Arts & Sciences assistant professor of computer science, is to harness the power of artificial intelligence. Their overarching objective is to develop AI-powered tools that can significantly enhance emergency care provided outside traditional hospital settings, ultimately leading to a greater number of lives saved, particularly among children. This initiative seeks to introduce a new paradigm where technology assists human responders in critical, time-sensitive situations. The AI models will be trained using comprehensive video recordings from the simulated emergency scenarios, allowing the technology to learn and identify optimal responses and interventions.
The research methodology relies heavily on meticulous data collection from realistic mannequin simulations. Boyle's team is gathering detailed video recordings of emergency responders addressing pediatric emergencies such as sudden heart or lung failure. This data encompasses critical metrics like the time taken for interventions, the precision of procedural skills, and the accuracy of medication selection and dosage. Following each simulation, responders participate in debriefing sessions, providing valuable feedback on how the tested technology, including videoconferencing, influenced their teamwork, communication, and overall efficiency. This comprehensive dataset will be instrumental in training advanced AI models, which will then be integrated into portable devices like phones or laptops, enabling emergency responders to leverage AI's analytical capabilities for rapid assessment and treatment. This approach addresses the current scarcity of real-life pediatric emergency video footage, which otherwise limits AI development in this critical domain, and circumvents patient privacy concerns.
Deepti Ghadiyaram elaborates on the transformative potential of the AI tool, illustrating its application in high-stakes pediatric emergencies. She envisions a scenario where EMS clinicians are performing cardiopulmonary resuscitation (CPR) on a child. The AI, equipped with audio and video analysis capabilities, would continuously monitor the responders' actions. It could then provide immediate, precise guidance, such as instructing them to adjust the speed or force of heart compressions, or reminding them to deliver breaths to the child. This real-time, dynamic feedback aims to optimize critical interventions, potentially surpassing human capacity for instantaneous assessment and correction. Furthermore, the researchers anticipate the AI's applicability extending beyond pediatric cases to adult emergencies, and even to personal smartphones, empowering parents or guardians with expert guidance when calling 911 for a child experiencing sudden cardiac or respiratory arrest.
Tehnaz Boyle highlights the unique difficulties faced by emergency medical service (EMS) clinicians when dealing with pediatric emergencies. Encounters with critically ill or injured children are statistically uncommon for most responders, leading to a lack of frequent practical experience. This infrequency makes it exceptionally challenging for them to recall and apply the specific, nuanced aspects of pediatric care when under immense pressure, at a moment when precision is paramount. The proposed AI tool aims to bridge this knowledge and experience gap by providing immediate access to specialized pediatric expert support in real-time, effectively expanding the EMS toolkit with a digital intervention. This is especially beneficial for individual responders working alone in complex situations, offering an invaluable layer of support and expertise that can significantly improve patient outcomes.
The ambitious research project has secured substantial financial backing, receiving a five-year grant totaling $3.7 million from the National Institutes of Health. This significant funding underscores the potential impact and scientific merit of the study. Boyle emphasizes that a key strength of this endeavor lies in its ability to provide an unprecedented, highly granular understanding of how care is currently delivered to critically ill children across diverse regions of the United States. Such detailed insights have never been systematically collected before. Ultimately, the researchers hope that the findings from this study will serve as compelling evidence, urging EMS systems, funding bodies, and government agencies to increase their investment in the development and implementation of digitally integrated prehospital systems. The goal is to fundamentally improve patient care and achieve better health outcomes for children in emergency situations nationwide.