Software developer using machine learning algorithms

August 19, 2022

Language recognition improves electronic care records documentation 

For an electronic care record to be successful, it must rely on quality data. Correctly entered and correctly correlated. To help documentation in homecare, Systematic has developed an explorative project that uses machine learning to simplify documentation.

Precise and easy documentation

In Thisted Municipality, Denmark, homecare workers can choose from a broad variety of categories when documenting their observations with patients in their electronic care record. The selection of categories is designed to produce highly accurate documentation and data, but in Thisted Municipality some homecare workers only made use of a handful of categories for their documentation. This led to inadequate handovers, missing documentation, and suboptimal workflows.

To help the homecare teams, Systematic has developed an AI-driven project together with Thisted Municipality to make documentation precise and easy. Building on Natural Language Technologies (NLP) across speech recognition, speech synthesis, and automatic form filling, the solution has demonstrated how machine learning can help improve documentation in Columna Cura Electronic Care Record.

Natural Language Technologies and machine learning

Documentation in one simple workflow

Following a technology exploration, Systematic’s Business Intelligence and Data Science team decided to focus on auto form filling to help solve the challenge. To simplify daily documentation, the auto form solution allows homecare workers to fill in all their documentation as free text. A combination of Natural Language Technologies and machine learning capabilities then analyse and classify the text and suggest the correct category for each individual observation.

Potential building blocks for future improvements

Learnings from the project indicate that speech recognition technology can potentially help elevate the solution even further. By using speech recognition, the solution could help homecare workers who struggle with grammar, either due to dyslexia or not having Danish as their first language.

The solution would not only help with correct documentation but could potentially help loosen language requirements for homecare workers in the future.