AI Implementeringsnätverk (AIDA-IN)

AIDAs nyaste projektform, AI-införandenätverk, syftar till att möjliggöra samverkan, kunskapsutbyte och gemensamt lärande mellan individer hos vårdgivare som arbetar med AI-implementering. Genom att kraftansamla påskyndas ett klokt och strukturerat kliniskt införande av AI-verktyg – en möjlighet att bidra till att Sverige tar en ledande position inom fältet och agerar föregångsland. Nedan beskrivs AIDAs 3 pågående nationella AIDA-IN projekt.

AIDA-IN: National network for the introduction of AI-assisted vertebral fracture screening (Osteoporosis) in clinical workflows (VertNet)

Anna Spångeus
Region Östergötland, LIU, Region Halland, Region Stockholm, VGR

This project aims to establish a national network to implement AI-based screening for vertebral fractures, a key indicator of osteoporosis. Osteoporotic fractures are highly prevalent, affecting one in two women and one in four men over 50, and are associated with significant morbidity, mortality, and healthcare costs. Vertebral fractures are the most common type and strongly predict future fractures, including hip fractures, yet fewer than one-third are currently diagnosed. Despite effective treatments that can reduce the risk of new fractures by up to 70%, most patients remain undetected and untreated.

AI-driven solutions for automatic detection of vertebral fractures in CT scans offer a transformative opportunity for early identification and secondary prevention. The network will support regions in Sweden that are implementing or planning such solutions, enabling knowledge exchange, harmonized workflows, and patient-centered processes. Future patient benefits include earlier diagnosis, timely treatment, and reduced risk of severe fractures, ultimately improving quality of life and lowering societal costs. The project also creates a scalable model for AI integration in other musculoskeletal conditions.

AIDA-IN: Clinical implementation of AI and tomosynthesis for breast cancer screening (AIBCS)

Fredrik Strand
Region Stockholm, Region Skåne, Region Östergötland, KI, LU, LIU

AI systems have been demonstrated to improve breast cancer screening in prospective trials and real-world implementations. Our network will further increase our understanding of how AI can improve current breast cancer screening through earlier detection of cancer, reduced recall of cancer-free women and a better use of available healthcare resources.

In two scientific studies, we will:

1. explore the impact of clinical implementation of AI to increase resources where AI is confident there are cancer signs and decrease resources where AI is confident there is no cancer.
2. explore the use of AI to make screening based on 3D-mammo feasible from a workload perspective

In addition, we will provide annual physical meetings inviting representatives of all Swedish healthcare providers to exchange experiences and ideas within our domain. The network is a collaboration between the universities Karolinska Institutet, Lund University and Linköping University, as well as the healthcare regions of Stockholm, Skåne and Östergötland.

AIDA-IN: AI-RAD Implement - Ett multiregionalt nätverk för klinisk AI-implementering

Michael Wilczek
Region Stockholm, Region Sörmland, Region Östergötland, LIU

Facing increasing examination volumes, specialist shortages, and pressure for shorter lead times, Swedish radiology might benefit from coordinated AI adoption strategies.

This project establishes a multiregional network for coordinated implementation of AI-based decision support in acute radiology. The collaboration between Södersjukhuset, Linköping University Hospital, and Nyköping Hospital will implement and evaluate two CE-marked AI algorithms: one for automatic fracture detection on skeletal X-rays and another for multiple pathologies in brain CT and thoracic imaging. The network aims to develop a scalable implementation model, create a handbook for commercial AI deployment, and establish routines for measuring AI performance and user acceptance. Physical meetings will be held annually, complemented by quarterly digital network sessions.

Future patient benefits include faster diagnostic workflows with reduced interpretation time, improved detection accuracy for fractures and critical findings, and more equitable diagnostic quality across different healthcare settings. Earlier detection of acute conditions will enable timelier treatment decisions, ultimately improving patient management.