More efficient and equal emergency care with advanced medical decision support tools
| Reference number | |
| Coordinator | Lunds universitet - Avdelningen för arbets- och miljömedicin |
| Funding from Vinnova | SEK 5 336 171 |
| Project duration | June 2018 - December 2022 |
| Status | Completed |
| Venture | Digital health |
| Call | 2017-04570-en |
Important results from the project
In the project, we have established an extensive research infrastructure with data from Region Skåne linked with national registers. The aim was to develop a coherent decision support tool for initial assessment of patients seeking emergency medical care. In several scientific publications, we have reported important building blocks of such a decision support tool and presented evaluations of the performance that is on the level with experienced physicians. We have also collaborated with Region Skåne regarding the implementation of decision support tools more generally.
Expected long term effects
We expect that a medical decision support tool for use in emergency healthcare can contribute to shortening processing times while maintaining a high level of patient safety, as well as equalizing differences in assessment depending on workload, time of day and staff experience. We also expect that our work can contribute to facilitating the introduction of medical decision support tools more generally in healthcare.
Approach and implementation
In the project, we have established an extensive research infrastructure with data from the Scania County Council in southern Sweden (Region Skåne) linked with national registers with secure data storage. We have developed applications using machine learning methods, such as artificial neural networks, and used Python and Tensorflow. We have also developed simpler statistical models using logistic regression. We have evaluated and compared the performance of the models in one or two distinct steps through cross-validations and separate external validations.