Leader: Margreth Grotle
Post doctor: Bjørnar Berg
PhD Students: Lise Grete Kjønø and Zheng An Toh
External Collaborators:
- Kjersti Storheim, FORMI, Division of Clinical Neuroscience, Oslo University Hospital
- John-Anker Zwart, FORMI, Division of Clinical Neuroscience, Oslo University Hospital
- Allan Abbott, Department of Health, Medicine and Caring Sciences, Linköping University
- Jan Hartvigsen, Department of Sports Science and Clinical Biomechanics, University of Southern Denmark
- Stine Clausen, Department of Sports Science and Clinical Biomechanics, University of Southern Denmark
- Tore Solberg, Department of Clinical Medicine, UiT-The Arctic University of Norway; Department of Neurosurgery, University Hospital of North Norway; The Norwegian Registry for Spine Surgery, University Hospital of North Norway
- Tor Ingebrigtsen, Department of Clinical Medicine, UiT-The Arctic University of Norway; Department of Neurosurgery, University Hospital of North Norway; The Norwegian Registry for Spine Surgery, University Hospital of North Norway
- Tarjei Rysstad, Dept. of Physiotherapy, OsloMet
- Karin Magnusson, Norwegian Institute of Public Health
- Nina Østerås, Division of Rheumatology and Research, Diakonhjemmet Hospital
- He Hong-Gu, Alice Lee Centre for Nursing Studies, Yong Lo Lin School of Medicine, National University of Singapore
- Thor Einar Holmgard, Norwegian Back Association
This project is funded by the Norwegian Research Council, and is the first external funded project in CIM.
The primary objective is to use machine learning methods on large survey and health register data to identify people with different treatment trajectories and health outcomes after surgical and/or conservative treatment for spinal disorders.
Secondary objectives are to 1) conduct external validation of the prediction models in data sets from Denmark and Sweden, and 2) explore how the prediction models can be implemented into AI-based clinical co-decision tools and interventions. The overarching aim of the AID-Spine project is to address health and welfare challenges in spinal disorders by aiming for a future personalized and sustainable healthcare.
Four major research questions will be addressed:
1) what characterizes patients with different treatment trajectories for spinal disorder(s) and who is at risk for receiving spine surgery as compared to conservative treatment?
2) what characterizes patients who achieve a minimal important change after conservative and/or surgical treatment?
3) how valid are these risk and prognostic models when tested in external data sets? and
4) how can these risk and prognostic models be implemented into personalized, meaningful AI-based clinical co-decision tools and/or interventions?
Our main hypothesis is that by using machine learning methods accurate risk and prognostic models for different health outcomes can be developed, and these will form the basis for AI-based clinical co-decision tools that facilitate implementation of a personalized and sustainable healthcare for people with spinal disorders.
The project is organized in three work packages (WPs), in which the two first research questions will be investigated in WP1, and the two last research questions in WP2 and WP3, respectively. Objective 1 will be addressed by using data from general population surveys (HUNT, Tromsø, and Ullensaker surveys) and administrative health registry data (Norwegian Patient Registry, NPR, and Norwegian Registry for Primary Health Care, NRPHC). By linking these registers we will explore risk profiles (based on a broad range of variables in the general population surveys) for receiving conservative and/or surgical treatment(s) for spinal disorders. Objective 2 will be addressed by using the merged data from surveys and administrative health registers used in objective 1, and link these to Norwegian spine registries. The analyses will be conducted separately for the two main treatment trajectories – conservative treatment and spine surgery. To study prognostic models for different health outcomes after conservative treatments data from the Norwegian registry for neck and back pain (NNRR) will be used. Similarly, prognostic models for different health outcomes after spine surgery data will be linked to the Norwegian registry for spine surgery (NORspine). Two core health outcomes will be measured: a minimal important change (MIC) in back-related disability measured by the Oswestry Disability Index, and unfavourable events following treatment (complications, re-operations).
In WP2 the risk and prognostic models will be validated in Danish and Swedish data sets (SpineData, Danish Spine Database (DaRD), DaneSpine and SweSpine).
In WP3 the validated risk and prognostic models will be integrated in clinical decision-making tools and tested in different clinical settings, e.g. in the first consultation between surgeon and patient referred for surgical assessment. The feasibility of implementing the clinical decision-making tools will be explored by qualitative interviews and observation in participatory co-design approach. The AID-Spine has received 3-years funding for two PhD fellows and one postdoc fellow. Prof Margreth Grotle is the PI with several strong project collaborators covering a broad interdisciplinary group, including neurosurgeons, physiotherapists, data scientists, epidemiologists, statisticians, a user group, and clinicians working with spinal disorders.