Period, duration
spring/summer 2025, 4 to 6 months
Profile required
BAc+3 or +4 student in Statistical Analysis, Machine Learning, Signal Processing, Matlab / Python (Pytorch), autonomous, curious, dynamic
Gratuity
4.35€/h (standard)
Location
Annecy-le-vieux ; Maison de la mécatronique ; SYMME and LISTIC laboratories
Subject
BACKGROUND :
Chronic obstructive respiratory pathologies, such as Chronic Obstructive Pulmonary Disease (COPD), represent a major public health issue. In this context, we are conducting a clinical study, called MUKROBS, in which clinical data are collected and adventitious breath sounds recorded at different sites in the chest of COPD patients. The aim of this study is to develop two congestion scores, one based on clinical indicators and the other on data obtained using a measurement device specifically developed for the study.
In the clinical study, thousands of raw acoustic recordings of different durations were collected from a pathological group and a control group, for a total of 60 subjects. Among these data, we aim to identify and count mainly 3 types of pathological respiratory sounds (crackling, sibilant, purring) with different temporal and frequency characteristics. Preliminary work has already been carried out on AI recognition of these sounds, with a view to obtaining indicators that will contribute to the construction of the second score. This work has raised a number of issues, particularly concerning the pre-processing of the measured signals.
MISSIONS :
The proposed internship will focus on two distinct areas:
A first line of work consists in consolidating the learning approach implemented in the previous work for AI-based noise recognition.
- The second line of work will be devoted to improving the AI models defined and developing new data augmentation strategies. A noise source localization approach will be considered as the project progresses.
The internship will be linked to another internship on the development of clinical scores based on the statistical processing of data recorded on patients, before and after their medical treatment.
Contact
Please send your transcripts for the last two years and your CV to :
Argheesh.Bhanot
@univ-smb.fr
Project team
Christine Barthod, teacher-researcher IUT Annecy / SYMME
Laurent Goujon, teacher-researcher IUT Annecy / SYMME
Argheesh Bhanot, teacher-researcher IUT Annecy / LISTIC