Stagiaire en étalonnage en science des données et apprentissage profond - Paris

💡 Produits ou services responsables
Stage
Localisation Paris
Pas de télétravail autorisé
Publiée le 01/04/2021

FeetMe

Depuis octobre 2013, FeetMe développe des dispositifs médicaux connectés pour améliorer la mobilité.

💡 Produits ou services responsables

La mission de cette entreprise est de concevoir des produits ou proposer des services éco-responsables alignés avec les besoins de la transformation écologique et solidaire.

Plus d'informations
Mesure d'impact
FeetMe n'a pas encore transmis de mesure d'impact
Labels et certifications
Cette structure n'a pas souhaité nous communiquer les labels ou certifications qu'elle a pu obtenir.
Voir plus

FeetMe est une jeune startup du domaine médical qui cherche à résoudre le problème des troubles de la mobilité́. Fort de notre expérience dans l’analyse de la marche, nous développons des solutions technologiques (objet connecté et applications) pour accompagner les patients dans le suivi et la gestion de leur maladie chronique avec des bio marqueurs numériques et des outils de suivi des études en vie réelle pour les patients, les médecins et les groupes pharmaceutiques.
FeetMe connait une forte croissance. Un premier produit d’aide au diagnostic pour les professionnels de santé a déjà̀ été́ commercialisé et FeetMe prépare le lancement de la V2 à destination des patients. L’équipe a remporté de nombreux concours et aides publiques (Concours Mondial de l’Innovation, Création Développement, FUI notamment) et a réalisé une levée de fonds lui permettant d’assurer son développement.


FeetMe is a digital health company developing innovative connected technologies and services to improve patient outcomes in functional care, track disease evolution, and optimize medication utilization.

The innovative technology from FeetMe allows gait and posture analysis in real-time and real-life conditions. The technology combines pressure sensors, motion sensors and learning algorithms to analyse patients’ functional capacity,as well as empower rehabilitation among sufferers of gait disorders.

FeetMe is growing quickly, its first product FeetMe Evaluation for diagnostics assistance for health professionals and for clinical research within pharma is on the market already. The company is preparing the launch of FeetMe Rehabilitation, a solution intended for home-based rehabilitation of patients suffering from walking difficulties.

It has formalized a collaboration with Novartis and AMGEN laboratories to improve the management of multiple sclerosis and osteoporosis.

Product

FeetMe Evaluation is a solution for ambulatory gait assessment that combines miniaturized pressure sensors, motion sensors and an embedded calculation power to allow real time gait parameters assessment

FeetMe Rehabilitation is a solution for home-based rehabilitation. It combines a choice of clinically validated exercises a pair connected insoles for real type movement measurement and an application that provides patients real time feedback

Project Description

The functional and motor skills of patients with neuropathologies are assessed using gait parameters.
They are key indicators for determining the progression of neurodegenerative disease. These walking
parameters are essential in the therapeutic decision making process. Currently, they are measured in
movement analysis laboratories and do not allow patient follow-up over time. There is therefore an
interest in evaluating them in real life. FeetMe insoles meet this first need by using inertial units and
pressur sensors. However, inertial units suffer from drift and bias that degraded gait metrics accuracy.
If such drift and bias can be corrected with classical statistical methods, we want to investigate new
methods based on deep learning.

The first part of the internship will consists in a literature review in the field. Then the intern will develop
several deep learning model for estimating drift and bias both on simulated and real data collected
with our insoles.

Finally, the intern will be in charge of comparing the models he has developed with the classical
approachs proposed in the literature.

Brossard, M., Bonnabel, S., & Barrau, A. (2020, June). Denoising IMU Gyroscopes with Deep Learning
for Open-Loop Attitude Estimation. In 2020 IEEE Robotics and Automation Letters. IEEE.


We are seeking a candidate with the following skills:
• Master’s Level Degree
• Strong interest and rigor in R&D
• Prior experience with deep learning
• Prior experience with a few of the following models: Logistic Regression, Linear Regression, Hidden
Markov Models, Conditional Random Fields
• Strong analytical and quantitative problem-solving skills.
• Proficient in one or more programming languages such as Python, MATLAB, R
• A drive to learn and master new technologies and techniques.
• Good oral and written communications skills to interact with other development and applications engineers daily