Apply

APPLICATION DEADLINE 21 August 2025

Sound Analysis for Predicting Category 1 Ambulance Calls 

• Supervisor: Dr Ning Ma  n.ma@sheffield.ac.uk  and Professor Jon Barker, University of Sheffield 
• Project Partner: Yorkshire Ambulance Service NHS Trust 

Your PhD will focus on the development of voice analysis technologies to enhance the prediction and triaging of Category 1 ambulance calls. 

Ambulance call centres play a critical role in triaging life-threatening medical emergencies. Category 1 calls, indicating life­ threatening injuries or illnesses such as cardiac arrest or severe respiratory distress, demand an immediate response to reduce avoidable fatalities. YAS handles over 1.1 million emergency and urgent calls to 999 annually. experienced call handlers often recognise the severity of cases within the initial 15-20 seconds of a call. However, the accuracy of these assessments can be influenced by factors such as the call handler’s expertise, call volumes, and stress levels, potentially delaying life-saving interventions. 

Emerging advancements in artificial intelligence (Al)-driven speech and voice analysis present transformative opportunities to enhance emergency call triaging. For instance, identifying specific audio features, such as laboured breathing or vocal markers of severe distress, could enable earlier and more accurate prediction of Category 1 emergencies. The integration of such tools into call centre workflows promises to improve decision-making speed and accuracy, ultimately saving lives. 

This collaborative PhD project aims to develop and evaluate advanced deep learning models for speech and audio analysis to predict Category 1 emergencies, improving the speed and precision of emergency response systems. The objectives include: 

• Collaborate with YAS to curate a high-quality dataset of emergency call recordings, annotated with corresponding medical outcomes and severity levels. 
• Identify vocal and acoustic biomarkers indicative of life-threatening conditions, including laboured breathing, distressed speech patterns, or cognitive impairment markers. 
• Develop machine learning models capable of predicting Category 1 emergencies based on real-time audio features extracted from calls. 
• Work iteratively with YAS researchers to test and refine the models, ensuring usability, reliability, and integration into operational workftows. 

The successful candidate will benefit from interdisciplinary training in experimental design, advanced speech analysis, and machine learning techniques. Supervision will be provided by experts from the University of Sheffield and industry professionals at YAS. The candidate will also undertake a placement at YAS to gain hands-on experience in real-world emergency call environments, ensuring the practical relevance and impact of the research. 

This research aligns with the Positive Uses of Sound theme in the Sound Futures CDT and addresses both national and international health priorities in developing fair and inclusive systems for real-world applications. 

Lead Supervisor https://sheffield.ac.uk/cs/people/academic/ning-ma