Discover, Connect, Collaborate: Find the Nordic teams working on similar AI use cases.
73 results
The program can measure lung lesions, detect blood clots, and outline areas of pathological lung tissue on CT scans. The goal is to support radiologists in their work with CT thorax (chest) examinations. Currently, international guidelines are not followed, which require tracking nodules using volumetric measurements. The program enables these consecutive measurements with automatic segmentation, plots growth and regression curves, and performs this across multiple scans taken at different time points. The work is concentration-intensive and repetitive, often taking a long time, as radiologists fear missing new (and potentially serious) lesions.
AI-based segmentation of the shoulder bone in CT data for automatic generation of a 3D model of the patient's shoulder joint for virtual surgical planning of artificial shoulder replacement surgery.
AI-based segmentation of the skull bones in CT data for automatic generation of a 3D model of the patient's skull for virtual surgical planning in corrective jaw surgery.
AI-based segmentation of coronary arteries in cardiac CT data for automated generation of 3D models of the patient's coronary arteries for 3D printing, aiding diagnosis, surgical simulation, and education.
Waiting
AI for pre-diagnostic and diagnostic assessment of prostate pathology. The system helps identify samples requiring further analysis and supports the evaluation of suspicious areas.
AI-RAPTOR is a secure, GDPR-compliant annotation platform designed for medical imaging data, particularly for identifying lung diseases in CT thorax scans. It integrates with PACS systems and supports the development, training, and validation of AI models using large image datasets. The platform enables annotation of various types of medical images and has potential for scaling across the region to free up workforce capacity and improve data quality.
The aim is to investigate the effectiveness of an AI solution for diagnostics in prostate, breast, and gastrointestinal pathology to assess how it can enhance diagnostic quality, consistency, and efficiency in diagnostic work.
AI algorithms are used to detect suspicious findings in mammography screening and clinical mammography. They support or partially replace the initial review in mammography screening to improve performance and save radiologist resources.
The AI uses semantic search and algorithms to efficiently extract and highlight relevant information from patient records during thrombophilia assessments. It improves information retrieval and ensures healthcare staff do not miss critical data. The tool reduces time spent reviewing medical records, freeing up staff capacity and increasing decision-making accuracy.
AI technology that plans, generates, adapts, and optimizes shift schedules
A clinical diagnostic pilot evaluation of an AI-based support tool for identification of mutations based on DNA sequencing.
The project aims to develop and test an AI-based Retrieval Augmented Generation (RAG) system – MIND (Medical Intelligent Note and Documentation Assistant) – that automates the creation of draft versions of interdisciplinary nursing communication documents based on data from electronic patient records. The system automatically retrieves and structures relevant information from the patient record, follows documentation guidelines, and generates drafts with traceable source references. A validation layer ensures factual accuracy, and a transparency report documents which record sections were used. The first proof of concept (POC 1) was completed with nurses from the Geriatric Department G and municipal nurses, who assessed the quality of the AI-generated care plans. Results show document quality comparable to nurses' current documentation, and the time required to produce care plans could potentially be reduced to less than half. Specific improvement points have been identified and will be implemented in the second proof of concept (POC 2). Funding for POC 2 is being sought within the near future.
Pilot testing of two AI-powered speech-to-note solutions that record, transcribe, and generate a draft clinical note based on clinical conversations. The solutions are being tested across nine different clinical settings in regional hospitals, in various conversation scenarios (ward rounds, outpatient visits, and acute consultations). The focus is on evaluating the solutions' potential for improving quality, saving time, user perspectives, and cost-effectiveness.
AI solution for pre-diagnostic assessment and decision support in pathology samples. It helps identify samples requiring additional special staining and supports cancer grading.
The AI initiative involves testing technology for assisted journal documentation during selected video consultations to assess the technology's maturity and the foundation for broader implementation. The project aims to improve documentation quality, save time, enhance patient experience, and increase staff satisfaction.
Pilot testing of technology for assisted medical record capture during selected video consultations to assess the technology's maturity and the foundation for broader implementation.
AIM-SAFE is a European project developing a secure and reliable AI framework for clinical decision support, with a focus on medication management and cancer treatment. The solution uses collaborative AI agents to predict side effects and optimize treatments, among other applications. It is designed with high data security, explainability, and full compliance with EU regulations. Federated learning enables knowledge sharing without transferring sensitive data. The goal is to improve patient safety, treatment quality, and efficiency in healthcare.
AI-Raptor performs screening of lung scans to ensure objectivity and to ensure the right treatment is assigned to the right patient.
AIRCON (AI Radiotherapy CONtouring) was developed as part of a quality improvement project in the Department of Medical Physics, Oncology Department at Sygehus Lillebælt, focusing on the integration of AI into cancer radiotherapy treatment. The goal is to develop AI models that support physicians during contouring for radiotherapy planning. The model performs a standard contouring, after which the result is sent for physician review, where necessary corrections are made to comply with Danish national guidelines. By using a model to generate the initial contours, we expect to achieve more consistent contouring, ensuring that all organs and tumor areas are delineated in the same way (reducing intra- and inter-observer variation), thereby increasing confidence in delivering correct and optimal treatment on the first attempt. Additionally, it is generally faster to make minor adjustments to an existing contour than to perform a full manual contouring from scratch, so we anticipate a reduction in the time required per patient for contouring.
We aim to develop an AI simulator that enables clinicians and students to train communication skills from the The Good Conversation (DGS) course using an AI-simulated patient via text and speech. The solution should be applicable both in course curricula and in daily clinical practice, allowing repeated practice of realistic scenarios in a safe environment with feedback based on DGS principles. Funding is sought for June 2026, but the project and development have not yet begun, except for an early prototype.
AI solution to support the diagnosis of lymph nodes in digital pathology. The solution aims to reduce the time spent on evaluating lymph nodes across many cancer areas.
Apollo Brain is an AI solution supporting stroke assessment, currently being tested in a pilot to evaluate whether younger radiologists using AI can achieve the same speed and accuracy as experienced neuroradiologists.
ARTHUR is a fully automated robotic system that scans both hands and wrists using ultrasound while guiding the patient through the procedure. All videos and images are sent to its AI, called DIANA, which scores rheumatoid arthritis activity on every image and generates a report that is transferred to EPJ South, so the physician/nurse has it available before the patient arrives.
ARTHUR robot for AI-supported ultrasound scanning of hands and wrists in joint gout. Used to ensure systematic and consistent scanning and to free up specialist physician resources.
AI-based evaluation of pharyngeal airway changes after corrective jaw surgery in CT data, for example to assess airway enlargement in sleep apnea patients treated with surgical advancement of the mandible or via chin advancement surgery.
TrueFidelity reduces CT image noise using deep learning image reconstruction, improving image quality, reducing radiation dose, and potentially decreasing the time radiologists spend reviewing images.
BoneXpert is used for automatic determination of bone age from hand X-rays in children. This helps improve the diagnosis of delayed or absent growth and optimizes treatment.
DANARC (Danish Centre for AI and Robotics in Cardiovascular Health) develops next-generation robotic technology for ultrasound scanning of the heart (echocardiography). The project combines robotics, artificial intelligence, and clinical expertise to create a system capable of performing standardized and reproducible heart scans with minimal operator dependency. The goal is to improve the quality and accessibility of echocardiography, support healthcare staff, and contribute to earlier and more accurate diagnosis of heart and vascular diseases.
Development of algorithms for fast and accurate search in patient records, reducing time spent reviewing records and laying the foundation for future AI projects.
DeSeRT 2 is an AI-based decision support system that helps detect critical conditions in acute patients by assessing the probability of life-threatening conditions. It analyzes biochemical tests, clinical assessments, and patient history to quickly determine disease risk and treatment needs.
Arterys Chest MSK AI identifies four types of findings on chest X-rays and assists younger doctors in the emergency department and primary care with avoiding missed important findings, especially pneumothorax.
Digital decision support for the diagnosis of porphyria based on biochemical tests. It supports clinicians in interpreting test results and provides more accurate and faster responses to porphyria investigations.
Syncsense VR solution that links training to natural and urban experiences through VR and sensors, to increase motivation, activity levels, and well-being among patients in care and rehabilitation.
Dora Research is an AI-driven assistant that helps identify and categorize relevant clinical information in medical records for rapid and consistent review for research purposes.
A non-invasive AI examination that complements the assessment of stable ischemic heart disease.
AI-based decision support to predict aggressive disease progression in prostate cancer. The goal is to distinguish between aggressive and less aggressive disease courses, thereby reducing unnecessary surgeries.
AI solution for real-time decision support during tissue sampling in cervical screening. The aim is to reduce the number of tissue samples from four to one and improve the diagnosis of cervical cancer.
Clinical decision support to prevent blood sample mix-ups and ensure correct sample identity across hospitals in the region.
AI solution for improving spinal treatment through data-driven evidence.
The AI algorithm NOAH automates the evaluation of retinal photos and classifies diabetic eye disease according to Danish guidelines. The solution is already in use at Odense University Hospital and Rigshospitalet and will later be implemented at additional clinics. Implementation is expected to reduce costs by 50-80%, free up ophthalmologist resources, and improve diagnostic accuracy, which is anticipated to reduce vision loss and blindness.
Clinical AI Guide is an AI chat partner for discussing all aspects of AI in the Region of Southern Denmark. Whether you are procuring, developing, or researching AI for the healthcare system, Clinical AI Guide prepares you for collaboration with clinics, the region, and industry partners. The solution is built on the Region of Southern Denmark's chatbot platform, SydChat, and draws on regional, national, and European sources for responsible procurement, development, and research in AI technology.
AI solution that estimates liver fibrosis stage from blood tests, used for early diagnosis of advanced fatty liver disease and prevention of cirrhosis.
MethoTrack is a daily automated monitoring system for patients on methotrexate (MTX) treatment. The system has been running in full clinical operation at the Rheumatology Outpatient Clinic, OUH Svendborg, since April 8, 2026. Developed within the department as part of a PhD project. The system automatically scans daily FMK prescriptions and identifies all MTX patients in the outpatient clinic. For each patient, it checks whether there is an active blood test request, whether the blood sample was taken on time, and whether the results show alarm findings (e.g., pancytopenia). Results are sent as Excel reports to each professional group in the clinic, containing only their respective tasks: - The secretary views patients missing requests and automatically sends standard letters when blood samples are not taken - The nurse calls patients who have not responded to the standard letters - The physician is alerted to alarm findings (e.g., pancytopenia) in blood test results Each team marks tasks as completed in the Excel file so they do not reappear the next day. The system captures blood samples taken outside the hospital (e.g., general practice, local laboratories) that would otherwise not appear in the hospital's inbox. It relieves all three professional groups of daily manual patient reviews and enables specialists to monitor patients remotely, even when patients are not physically present for follow-up. In one year of operation, the system identified three patients with MTX-induced pancytopenia. The most recent case was detected solely by the system, as the blood tests were ordered and taken in general practice and thus never reached the hospital's inbox.
In this project, we are developing an AI-based solution that can automatically extract and assign the correct ICD diagnosis codes from free text in electronic patient records.
Digital solution for the treatment of patients with depression, supporting individualized digital treatment through generative AI.
An AI-based system that predicts patient arrivals and supports staffing in emergency departments. Short-term forecasts predict congestion hour by hour, while long-term forecasts predict congestion months ahead. The system is currently used at Odense University Hospital and is planned for implementation at Hillerød Hospital. It enables improved scheduling, reduced unnecessary costs, and better working conditions.
The project's overarching goal is to develop and test a prompt-chain system consisting of three collaborating agents – a writing agent and two validation agents – capable of converting PET/CT image reports into patient-friendly text. The project focuses on patients with metastatic breast cancer and on PET/CT as a nuclear medicine examination.
AI-based delineation automatically outlines risk organs on CT/MR scans for radiation therapy. This reduces the time required for organ contouring and ensures more consistent results across cancer departments.
It has been politically decided that all hospitals in the region will trial RBfracture from Radiobotics over a three-year period. Licensing and administrative costs are covered by Regional IT. The license runs from 01.02.2026 for a duration of three years, effective from now. The algorithm will be deployed directly on the imaging modality, i.e., as software in the X-ray rooms, with images sent to PACS showing the AI algorithm's findings as an overlay. The tool will primarily be used by the Emergency Department and orthopedic surgery in acute patient situations, but as a radiology department, we will receive the images for reporting and thus contribute to evaluating the algorithm's suggestions. Lillebælt Hospital has used the algorithm on a trial basis for the past couple of years, and fortunately, we can leverage their experience with both the technology and collaboration to move forward. We are next in line for implementation after Sygehus Sønderjylland, at Esbjerg and Grindsted Hospitals. OUH will follow at a later stage.
System for preventing falls to monitor and prevent patient falls.
The Visiopharm-metastasis app AI tool is used to detect metastases in sentinel lymph nodes in breast cancer. It supports pathologists in their decision-making during screening and may reduce time consumption in the long term.
The aim of the project is to develop an AI-based solution to assist pathologists in SNOMED coding of tissue samples. Particularly with large, coding-intensive preparations—such as in tumor diagnostics—coding is time-consuming and complex. The project will investigate how artificial intelligence can suggest relevant SNOMED codes (topography, morphology, procedures, etc.) based on the pathologist's descriptive text and clinical information.
An AI model for diagnosis and outcome prediction in intracranial aneurysms, supporting diagnosis and planning for unruptured aneurysms.
With "The Precision-Recruit Project", our goal is to screen all cancer patients based on their individual disease and offer them participation in the most recent available treatment. Through this project, we aim to develop, test, and implement an AI-driven platform that can automatically analyze patients' electronic health records to quickly and accurately identify and guide those who might be eligible for active research projects based on the protocol's specific inclusion/exclusion criteria.
The model predicts cancer risk within the next three months based on blood samples and is used for early detection of cancer in patients with increased risk.
Clinical decision support tool that can assist physicians in prescribing antibiotics
Using small language models and potentially machine learning models, we aim to analyze blood samples and electronic health record data to investigate the value of follow-up visits on the neurology ward. Initially, the focus is on epilepsy patients. Funding has just been approved, and we will launch the project in August 2026.
As part of a PhD project, a system for quality assurance of AI bone fracture algorithms is being developed and tested. The system is designed to read image descriptions, extract diagnoses from them, and compare these with the findings of the bone fracture algorithm. We are testing how accurately the system can perform comparisons, in order to develop a tool for automated quality assurance.
External validation of an MDR-approved AI fracture detection algorithm and theoretical testing of time-saving elements with a revised workflow where AI serves as a potential supportive tool for radiologists.
The AICE project aims to create a complete and validated AI-assisted pathway that improves CCE diagnostics, making the technology clinically viable and beneficial for patients, healthcare systems, and society. The project focuses on finalizing and externally validating AI algorithms, establishing a clinical support system for secure data handling, storage, and transmission, and developing a diagnostic pathway that integrates quality, efficiency, patient preferences, ethics, and economics. AICE will also support implementation in clinical practice through guidelines and scalability adjustments.
CHOICE develops an AI-based decision support solution that can automatically identify individuals at high risk of osteoporotic fractures using existing health register data. The aim is to support early detection, targeted DXA assessment, and preventive treatment in primary care
DETECT-AI is a research and development project in Region Syddanmark aiming to develop and test artificial intelligence for automated assessment of coronary calcification on routine chest CT scans. The project seeks to identify signs of subclinical atherosclerosis in patients who otherwise would not systematically undergo such evaluation. The goal is that AI-based analysis of already performed CT scans could contribute to earlier detection of patients at increased risk of cardiovascular disease and support more targeted preventive treatment. The project is carried out in collaboration between clinical departments, research environments, and technical partners, with focus on algorithm development, clinical applicability, and future implementation within the healthcare system.
This project develops a technical evaluation setup that enables systematic measurement of accuracy, completeness, and safety of AI-generated summaries. The focus is on objective performance metrics across all specialties.
Used as a supplement to filling out meeting minutes during staff meetings.
In this project, we aim to develop a prototype of a digital tool that automatically collects and organizes concise disease summaries for gout patients based on data from their electronic health records.
Remote environments (e.g., ships and platforms) often operate without a physician or nurse physically present. Additionally, language barriers and cultural differences make the collection of medical information difficult. Marina Health is a medical device software that assists non-physicians in collecting and documenting medical information and overcoming language barriers. This information is then forwarded as a medical report to the responsible physician, potentially providing more comprehensive information to the physician.
The project has developed an AI model for patient complaints. The model categorizes patient complaints according to the Healthcare Complaint Analysis Tool (HCAT). Currently, this is done manually, which is resource-intensive and leads to delays in response. Automated coding enables timely coding relative to the time of complaint receipt, thus providing a basis for improved learning. The model is ready for implementation. Developed together with Lars Morsø and Søren Bie Bogh, OPEN
Automatic generation of scan reports for PET/CT scans in metastatic breast cancer. Additionally, use of low-dose scans for prediction of diagnostic-dose scan results.
PROTECT is an AI-based research project in digital pathology focusing on the quantitative analysis of pancreatic cancer in digitized histological sections. The project combines pathological expertise from OUH with AI and computational expertise from SDU/IMADA. Using machine learning and a 'pathologist-in-the-loop' principle, models for quantitative histopathological analysis are developed and validated, with the aim of future integration into clinical digital pathology workflows. The goal is to improve reproducibility, standardization, and quantification of pathological assessments.
Automated note-taking for administrative meetings
The HR department at SLB uses a Chatbot to assist the payroll team and SLB employees with quick access to information from the personnel handbook and selected external websites
Can AI help differentiate between benign and malignant kidney tumors under 4 cm? This would greatly benefit patients by potentially avoiding biopsy. The collaboration will involve urology and radiology departments, CAIX, Pathology, and Nordic urology departments in Norway, Sweden, and Finland.