Smart Inhaler

Health & IoT

Background

One of the country’s pharmaceutical industry leaders developed an innovative inhaled drug. Due to the strict legal regulations, whole treatment has to be monitored which requires a customized smart inhaler that can be used to control its usage and send data to the system.

The primary demanded functionality was the ability to record the patient’s breath during medicine intake. For drug therapy, it was crucial to assess the correctness of the intake right after the patient’s inhalation to let the doctor monitor the treatment process.

Project Goals

  • Collect the sample recordings in the laboratory, that can be used to create the model prototyp.
  • Create the prototype of the algorithm that can extract features from the recorded sound and assess if the medicine was taken correctly.
  • The algorithm has to be run in the chip inside the inhaler.

Challenges

  • The algorithm had to be robust – and the inhaler can be used by in diverse environments.
  • The relation between the airflow sound and the quality of the intake may not be clear.
  • The computational resources in the inhaler’s chip are constrained.

Our work

When the patient is taking a dosage the sound of the inspiration is recorded and provided to the algorithm. The algorithm detects the exact moment of the breathe and cut out all of the other sounds.

Afterward, multiple different features are extracted from the sound. Using those features the Machine Learning algorithms can assess the quality of the intake and estimate the parameters of the patient’s breath. The data are stored in the cloud and the doctor can monitor the therapy online using the desktop app.

The outcome

Benefits

  • Enhanced safety issues at the workplace.
  • The shift manager is immediately informed about the disturbances in the workflow.
  • Faster, a more directed reaction in case of an unwanted event.
  • No need of a person watching all the CCTV streams all the time – people can allocate their time to other, more creative tasks.
Technology used
  • Python
  • Matlab
  • Librosa
  • Scikit-learn
Client
Schedule a call with our team