Andreas Triantafyllopoulos

Andreas Triantafyllopoulos

AI for Speech & Health

Speech for respiratory health

Broader context

The quintessential symbol of a doctor is the stethoscope. This is because auscultation offers us an unobtrusive way of monitoring the human body. Nowhere is this more important than when it comes to diseases that affect the upper or lower respiratory tract.

We aim to make the smartphone the modern-era equivalent of the stethoscope. By analyzing speech, we hope to identify whether a person is suffering from a respiratory disease, ranging from short-term infections (COVID-19, flu) to long-term conditions (asthma, COPD).

Current status

The majority of our work is done on a small-scale COPD dataset collected at the University of Augsburg (see references below). We have obtained promising results using interpretable features as input to our models. Currently, we are trying to improve the predictive performance of our models by investigating deep learning architectures, as well as expanding our scope by considering additional datasets and diseases.

Open research questions

The main open research question in respiratory health are:

  • whether we can distinguish between diseases (and not simply whether a person has a disease or is healthy)
  • whether speech can detect changes in disease status
  • whether we can make models that generalize across different languages

References

Triantafyllopoulos, A., Fendler, M., Batliner, A., Gerczuk, M., Amiriparian, S., Berghaus, T. M., & Schuller, B. W. (2022). Distinguishing between pre-and post-treatment in the speech of patients with chronic obstructive pulmonary disease. arXiv preprint arXiv:2207.12784.

Triantafyllopoulos, A., Batliner, A., Mayr, W., Fendler, M., Pokorny, F., Gerczuk, M., … & Schuller, B. (2024). Sustained vowels for pre-vs post-treatment COPD classification. arXiv preprint arXiv:2406.06355.

Mayr, W., Triantafyllopoulos, A., Batliner, A., Schuller, B. W., & Berghaus, T. M. (2025). Assessing the clinical and functional status of COPD patients using speech analysis during and after exacerbation. International Journal of Chronic Obstructive Pulmonary Disease, 137-147.