Facebook Pixel

Paid project - optimizing machine learning engine for MedTech startup


Company description

Sarita CareTech is a Danish MedTech startup with a 5th generation technology to make life easier for elderly people. Our strategy is to build a company attuned to the market needs bringing a new innovative product to the care industry with service design and development as the driving force. A product where care of the elderly is uniquely adapted to the person.

We currently develop a wearable brooch which is worn by elderly which has a fall detection feature. This allows the unit to alarm relevant parties if the user has fallen by automatically detecting the fall. 

Project description

Our product is able to automatically detect if a user that wears our product has fallen. This is done via data processing of an accelerometer and gyroscope. The current algorithm for detecting the falls consists of several variables which has to be tuned in order to achieve the best fall detection ratio. All the processing is currently happening only on the accelerometer data however gyroscope data is also sampled and stored but not used for anything yet.

Our challenge is twofold:

  • To be able to tune the fall detection algorithm parameters such it will have the best performance.
  • To be able to use the gyroscope data to assist in the fall detection process.

We want to utilize machine learning in order to increase the success rate of our fall detection system and also decrease the amount of false negatives.

The goal of the project is to implement machine learning system which constantly improves the fall detection algorithm as more user data has been gathered. We would also like to add more data from different sensors to evaluate if the system performance will increase. More information on technical part will be provided for applicants

Student description

This is a paid project to which we seek for a talented Master Thesis student or recent graduate to join our team for 3-6 months. We expect you to be able to do programming such that a solution for our company can be created. This means that a MatLab simulation to show the principles is okay for proof of concept yet not an adequate final solution. The final result must be a program written in C#/JAVA/Python where the source code is accessible for the company.The student/graduate is required to have a mid/advanced knowledge about machine learning and a good understanding of how Inertia Measurement Systems (Accelerometers Gyros and so on) work.

You will have access to our fall database currently containing around 200 falls sampled by accelerometer and gyroscope. We will also expect you to be able to collaborate with two high level software developers from the company and have weekly personal or online strategy meetings. For this project you dont need to be located in Denmark.

All the rights of the final solution has to belong to the company.

Please send us your 150 words application via online form below attaching your CV. Applications are being screened on an ongoing basis where the latest date to apply is 30th November. The application form will be closed right after the match is found or deadline expired. For questions or more information please contact us at info@matchmythesis.com 

Paid project - optimizing machine learning engine for MedTech startup | Match My Thesis
Feb 04, 2017


Machine learning


Data science

Deep learning

Mathematical modeling