The Omda Incident business unit provides robust systems for managing every aspect of Emergency Response. The Incident product solution supports case management, GIS (geographical information and maps), dispatching of resources and communication between all parties involved in the emergency response chain. It is deployed in control rooms, e.g. for 112 organizations.

Every year, the Swedish national emergency number (112) handles 3.5 million calls. Out of these, 650 000 are categorized as “silent calls”. While any of these could be real emergencies, current technical solutions and processes make manpower a limiting factor and thus the call must be ended to prioritize other incoming calls.

Most of the silent calls are not real emergencies (e.g. pocket dials / butt calls) but do still occupy the operators unnecessarily. Early classifying of these calls could shave off invaluable seconds which could mean the life or death in an actual emergency call.
This thesis aims to alleviate these problems by investigating ML/AI solutions which may provide means for helping operators to make faster and more accurate decisions and in turn save more lives.

Master thesis objective:

The objective of this thesis is to develop an ML/AI model that can 1) detect silent calls that represent real emergency calls; and 2) provide an early classification of silent calls. The solution will be assessed according to relevant business and technical KPIs.

Our tech stack:

• The model will be developed using suitable tools (e.g. Tensorflow) and languages (e.g. Python), to train, tune and validate the model.

What we offer:

• Access to data and development environment
• Mentor and Omda employees' domain knowledge

Competence requirements:

• Data Science (AI/ML theories and models)
• Experience in programming in Python
• Experience with machine-learning and AI projects
• Self-going and problem-solving
• Language: proficient in both Swedish and English
• Swedish Citizenship

Looking forward to meeting you!