Decision Support in Health Care using Artificial Intelligence (AI)

Goals of the Project

• To collect data from medical computer systems from both sides of the border in order to establish a strong database for this project and future projects.  The data will be deidentified and structured so that it can be easily analysed with machine learning methods.
• To develop decision support systems for a selection of services within the health care system based on new and known machine learning methods. It is important that new systems are thoroughly tested and validated under real-life conditions and that these systems, that are developed, can be used in accordance with the GDPR.

Areas of Research Focus

Datalabb

• There are multiple ongoing projects from which large databases are being developed. From these databases and with the support from AI, healthcare providers will have the opportunity to decide on the most effective care plans for patients. For example, one project has created a database from 1600 patients for the prediction of thrombosis or blood clot. An additional database from 5000 patients was created to predict the risk of recurrence of both thrombosis and bleeding.  These data have been anonymised and thus are available to other research institutions upon request. “Through the thrombosis projects, we have developed good routines for collecting and quality assuring the data,” says project leader Lars V. Magnusson.
• Together, we are working on the establishment of a database that will support ambulance staff with deciding on the most suitable hospital location for a patient given his/her health condition. 
• Another project in development is the creation of two datasets, one in Norway and one in Sweden, which will provide improved treatment support for trauma patients.  As Lars V. Magnusson mentions, “We have just started with the analysis of the data, so it is too early to draw any conclusion in relation to routines and strategies for using machine learning (AI) in this domain.”

Trauma

Quality in prehospital services; Observations, assessments, triage and interventions in ambulance services and the effect on the trauma patient’s outcome.

• Background – Trauma is the most common cause of death for people under the age of 45.  Good management of trauma patients commences when the need for care arises (prehospital care) until the patient reaches the right medical care destination. Efficient and effective care management is of the utmost importance to reduce the number of patient deaths. 
• Project Participants
  • The sub-project is carried out as a collaboration between Østfold University College’s Faculty of Health, Welfare and Organization and Faculty of Information Technology, Engineering and Economics, Østfold Hospital Trust, National Competence Center for Prehospital Acute Medicine (NAKOS) and Borås University.
• Objective
  • To explore the quality of the trauma patient's care management course through the health service by:
    • mapping prehospital observations, assessments, priorities and interventions
    • exploring correlations between prehospital staff competence, time of day the trauma occurs, and prehospital patient management (including observations and interventions)
    • mapping characteristics of the trauma patient, such as Rapid Emergency Triage and Treatment System (RETTS) score, New Injury Severity Score (NISS), characteristics of ambulance transport (transported to; stay on site, to GP, to emergency room, to hospital), time aspect (time to event location, time on site, transport time) and final diagnosis
    • comparing the patient's RETTS score versus the correct RETTS score based on literature; and
    • predicting which patients are at risk of developing complications after the experienced trauma based on the data and AI analysis; and using this data in the decision-making process when selecting the patient’s treatment plan (transport to general practitioner (GP), transport to emergency room, transport to hospital, received by the trauma team or not).
• Data Collection
  • Data is obtained both from data registered at the hospital and from the local trauma registry.  In our analysis, we will compare the national differences in patient outcomes between Norway and Sweden.

Deep Vein Thrombosis (DVT)

• Background – Annually in Østfold, there are approximately 1000 patients referred to the hospital with possible deep vein thrombosis (DVT).  Of those patients in question, positive diagnosis is confirmed in 1 in 5 patients. These numbers are also applicable for those questioning a pulmonary embolism (PE) as well.
• Objective – To research the primary care providers’ experiences with patients with DVT and to improve decision support system for DVT with help from AI. In addition, we have a survey to map the decision/treatment tools used when DVT and/or Pulmonary Embolism (PE) are suspected.
• Goals
  • Map the current tools, which are used in the decision-making process with respect to symptomatic patients, and the objective signs of DVT in the municipality, ambulance and specialist services on both the Norwegian and Swedish sides.
  • Develop and improve decision-making tools
• Activities
  • Collaborate with the Østfold Hospital Trust with regards to available data material that can be used with decision-making tools and improve these tools in conjunction with the departments of health and welfare and of information (machine learning group) at Østfold  University College.
• Anticipated Results
  • The results from the planned studies are expected to provide key information about the need for decision support based on artificial intelligence (AI) when considering both the treatment plans for patients with DVT and the perspectives from both healthcare personnel and the patient.

Main Partner Group and Financial Support 

All collaborating project partners are also members of Interreg Sverige-Norge, the European Regional Development Fund.  The project is partially supported with Interreg funding, which will last until 2022. 

The Project Partners with Interreg Sverige-Norge

• Østfold University College
• Østfold Hospital Trust
• Europeiska Unionen
• Borås University
• Aweria- Digitala system och beslutsstöd för akutsjukvård
• Chalmers University of Technology
• eSmart Systems
• Medfield Diagnostics
• Institutt for Energiteknikk
• Västra Götalandsregionen
• Lindholmen AI
• IFE- Institute for Energy Technology
• PICTA- Prehospital ICT Arena
• Ambulansesjukvården i Västra Götalandsregionen
• Sarpsborg Municipality
• Swedetraum
• PreHospen- Centre for Prehospital Research