I was very excited to be able to present to PhD students at the University of Sheffield from the Speech and Language Technologies CDT. In this presentation, I discussed the issues facing the use of AI in real-world medical settings and how these could be managed. I also discussed the current data collection being carried out in Leeds Teaching Hospitals NHS Trust. It was great to be able to connect with other researchers working in the field of speech and language technologies and hear their thoughts and suggestions for my future research.

In May, I had the pleasure of presenting at the annual UKRI AI CDTs in Healthcare Conference hosted at the University of Edinburgh. At this conference, PhD students from the University of Leeds, the University of Edinburgh, Imperial College London, and University College London, all studying the use of AI in healthcare, gather to discuss ideas. I had the honour of being chosen to present my work on behalf of the University of Leeds. In this presentation, I discussed the issues of making models that can be used in real-world scenarios, as well as the data we are collecting from Leeds Teaching Hospitals Trust.

In 2023, we published at Interspeech, which is an international conference on speech and language technology. Mary attended this conference and presented her work to other speech and language technology researchers. See below for a description of the paper:

We had so much fun at Be Curious last year that we went again this year! We showed about 300 people (mainly families with primary school aged children) just how amazing machine learning can be. We challenged our system to guess how old people were from how their voice sounded and put families to the test seeing if they could beat the computer in guessing the age and gender of people based only on their voice. We love being able to share our work with so many people and inspiring the next generation of scientists and researchers.

I attended the Leeds Cancer Research Centre celebrates World Cancer Day event and presented a poster of my work. Several researchers and clinitians presented their current cancer research with a focus on closing the care gap.

In November of 2022 I attended the Women in Machine Learning workshop in New Orleans. At this workshop I presented a poster summarising my current work in the detection of voice disease using artificial intelligence. This work used two different methods to classify patients with voice diseases from healthy controls from recordings of the patient’s speech. The two methods were 75.5% and 69.8% accurate when classifying new data. In order to simulate a real environment these new speech recordings were overlayed with backgroud noise and then put into the classifier again. The accuracy of the classifiers dropped to 66.0% and 62.3% respectively. By adding backgroud noise to the recordings the accuracy of the classfiers decreased and were more likely to classify healthy patients as having a voice disease.

Be Curious is an event run by the University of Leeds aiming to show the public some of the research happening at the University. We presented our work at Be Curious under the name “What can computers learn from your voice?”. At the event we showed families how computers can be used to learn things about your voice. We created an app which could predict a person’s age based on their voice. We also challenged people to beat the computer in guessing the age and gender of a person based only on their voice.