AIAnders Aasted Isaksen

Anders Aasted Isaksen

*This is an algorithmically generated author record based on publicly available information.

Current Organizations / Institutes

The organizations/institutes from the scholar's latest article

Previous Organizations / Institutes

The organizations / institutes from the scholar's other articles in the past ten years


Anders Aasted Isaksen has a total of 39 co-authors. Here are the top co-authors by number of shared publications:


Latest Publications

  • Open Access

    Potentially inappropriate prescribing in individuals with type 2 diabetes mellitus: a Danish nationwide study

    by Anne Sofie Baymler Lundberg, Anders Aasted Isaksen, Annelli Sandbæk, Anders Prior

    Aim: We aim to describe the prevalence of potentially inappropriate prescribing (PIP) in individuals with type 2 diabetes mellitus. Furthermore, we aim to explore associations between PIP and different individual-level factors. Subject and methods: This was a nationwide register-based cross-sectional study. The study population consisted of all Danish citizens with type 2 diabetes as of 1 January 2022, and who were diagnosed at 40 years of age or older. Individual-level factors were sociodemographic variables, disease burden, and contacts with health care providers. The outcomes were PIP categorised as general or diabetes-related. General outcomes were assessed using the Screening Tool of Older Person’s Prescriptions and Screening Tool to Alert Doctors to Right Treatment (STOPP/START) criteria, and diabetes-related outcomes were defined according to selected guideline recommendations. Results: We found that the prevalence of a general PIP omission affected nearly one third of the population. The most common type of under-prescribing was related to cardiovascular diseases. Similarly, the most prevalent of the four selected diabetes-related prescriptions was the non-initiation of recommended cardioprotective medication among individuals with type 2 diabetes and cardiovascular disease. The individual-level factors associated with PIP were female sex and living alone. Conclusion: Future initiatives could focus on individuals with type 2 diabetes and cardiovascular diseases and consider the individual-level factors associated with increased risk to guide targeted interventions.

  • Open Access

    Foundation model–derived representations of continuous glucose monitoring data: relation to glycemic patterns and cardiovascular risk (Preprint)

    by Helene Bei Thomsen, Signe Toft Andersen, Guy Fagherazzi, Anders Aasted Isaksen, Adam Hulman

    BACKGROUND Background: Foundation models can map continuous glucose monitoring (CGM) data into learned glucose representations, offering new potential for cardiovascular disease (CVD) risk prediction. OBJECTIVE This study investigated the glucose patterns captured by a foundation model and evaluated whether CGM data can improve the prediction of elevated N-terminal pro-BNP (NT-proBNP), a heart failure biomarker METHODS Method: Two datasets, including either people with type 1 diabetes (n=205) or type 2 diabetes (n=159), were analyzed. Representations derived from CGMformer were evaluated using uniform manifold approximation and projection (UMAP) and pairwise cosine distance analysis. Logistic regression with bootstrap optimism correction predicted elevated NT-proBNP (>125pg/mL) using clinical risk factors, the SCORE2-Diabetes score, CGM-derived metrics, and learned representations. RESULTS Results: Cosine distance and UMAP analyses demonstrated the foundation model captured individual-specific glycemic patterns (HbA1c and time-in-range(TIR)). UMAP dimensions for CANCAN dataset showed strong correlations with TIR (ρ=-0.71 to 0.81, p<0.001) and HbA1c (ρ=0.48 to -0.65, p<0.001), whereas correlations with NT-proBNP were weak (ρ=-0.04, p=0.063 and ρ=0.18, p<0.001). Similar and statistical significant (p<0.001) but weaker correlations with TIR and HbA1c were observed in T1DX. Prediction models with clinical predictors or SCORE2-Diabetes score achieved optimism-corrected area under the receiver operating characteristic curve (AUROC) of 0.79 [95%CI:0.76-0.87] and 0.74 [0.65-0.81] respectively. Adding the CGM-derived metrics time-in-range and mean amplitude of glycemic excursions achieved AUROCs of 0.81 [0.77-0.88] and 0.77 [0.70-0.85], replacing them with CGMformer based representations resulted in AUROCs to 0.79 [0.75-0.87] and 0.72 [0.65-0.81], and increased overfitting. The UMAP dimensions from the CANCAN data were strongly correlated with TIR (UMAP dimension 1: ρ=-0.71, p<0.001; UMAP dimension 2: ρ=0.81, p<0.001), and moderately correlated with HbA1c (UMAP dimension 1 (UMAP 1)ρ=0.48, p<0.001; UMAP dimension 2(UMAP 2): ρ=-0.65, p<0.001), while only weak correlations were observed with NT-proBNP (UMAP 1: ρ=-0.04, p=0.063; UMAP 2: ρ=0.18, p<0.001). Weaker correlations were observed for the T1DX dataset for both TIR (UMAP 1: ρ=-0.39, p<0.001; UMAP 2: ρ=-0.65,p<0.001) and HbA1c (UMAP 1: ρ=0.30, p<0.001; UMAP 2: ρ=0.50,p<0.001). CONCLUSIONS Conclusions: Learned representations capture both individual-level glucose patterns and meaningful structures when mapped against HbA1c and TIR. No patterns were observed upon inspection for CVD-related markers and CGM data did not improve prediction of NT-proBNP beyond established clinical variables. Larger prospective studies are warranted to examine whether individual glucose patterns may improve CVD prediction in diabetes

  • Open Access

    Influenza Vaccination Among Individuals With Type 1 Diabetes in Denmark: Coverage, Persistence, and Determinants, 2015–2022

    by Vithiya Karunakaran, Ida Borreby Pedersen, Astrid Hjelholt, Mads Kjølby, Esben Thyssen Vestergaard, Ole Frøbert, Anders Aasted Isaksen, Kurt Kristensen

    Aims: To investigate influenza vaccination coverage (IVC) and persistence with annual influenza vaccination among individuals with type 1 diabetes in Denmark and to identify clinical, demographic, and socioeconomic factors associated with vaccination status. Materials and Methods: We conducted a nationwide, register‐based cohort study of individuals aged 1–64 years with type 1 diabetes across the influenza seasons 2015–2022 in Denmark. Annual influenza vaccination status was assessed, and the cohort was stratified into age groups (1–18 and 19–64 years) according to age‐specific vaccine recommendations. We investigated associations between non‐vaccination/non‐persistence and demographic, clinical, and socioeconomic factors using Poisson regression to compute adjusted relative risk (aRR). Persistence was defined as receipt of influenza vaccination in more than 80% of eligible seasons during follow‐up. Results: We identified 25 473 individuals with type 1 diabetes corresponding to 168 212 influenza seasons. In 2015/16, IVC was 5.6% in children and 18.3% in adults. IVC remained stable until a marked increase in 2021/22 (15.7% and 41.8%, respectively). In children, longer diabetes duration, HbA1c ≥ 53 mmol/mol, and residence in The North Denmark Region were associated with non‐vaccination. In adults, male sex, younger age, shorter educational attainment, and HbA1c ≥ 53 mmol/mol were associated with non‐vaccination. In both populations, factors associated with non‐persistence were similar to those observed with non‐vaccination. Conclusions: In Denmark, IVC among individuals with type 1 diabetes remains well below WHO targets, highlighting the need for targeted vaccination strategies. Sociodemographic and clinical factors associated with non‐vaccination identify population groups where preventive efforts should be prioritised.

  • Open Access

    Continuous Glucose Monitoring–Derived Metrics and Cardiovascular Risk Among People With Diabetes: Systematic Scoping Review

    by Helene Bei Thomsen, Benjamin Lebiecka-Johansen, Ole Nørgaard, Tue Helms Andersen, Signe Toft Andersen, Guy Fagherazzi, Adam Hulman, Anders Aasted Isaksen

    Background: Conventional clinical markers guide cardiovascular risk stratification; however, continuous glucose monitoring (CGM) data remain absent from prediction models. A synthesis of the current literature is needed to clarify the prognostic relevance of CGM data for cardiovascular outcomes in people with diabetes. Objective: This scoping review aimed to identify published studies examining (1) the associations between glycemic control and cardiovascular outcomes and (2) the predictive value of CGM-derived metrics in cardiovascular risk assessment. Methods: MEDLINE and Embase were searched from inception to March 11, 2025, for peer-reviewed, original research that included CGM-derived metrics and cardiovascular disease (CVD) outcomes. Two reviewers screened the records independently. Results: A total of 53 studies were identified. These studies focused on type 1 diabetes, type 2 diabetes, both diabetes types, or prediabetes. Clinical outcomes were examined in 16 studies, while subclinical outcomes were assessed in 40 studies. Of the 53 studies, 47 were cross-sectional studies and 6 were longitudinal studies. All studies were association studies, and 3 included secondary analyses of predictive performance. However, none applied machine learning–based methods. A wide range of CGM-derived metrics and CVD outcomes, both clinical and subclinical, were studied in the literature. Conclusions: Overall, the findings were inconsistent across studies, and this was likely due to methodological weaknesses such as underpowered analyses. Time-in-range was both the most studied metric and associated with cardiovascular risk in the largest single study. Only the mean amplitude of glycemic excursions was consistently associated with CVD in most studies investigating this metric, when using statistical significance as a pragmatic indicator of consistency across heterogeneous studies. The prognostic value of CGM-derived metrics for CVD outcomes is currently underexplored. Longitudinal prediction studies on clinical CVD outcomes, leveraging the potential of routinely collected CGM data, are needed.


  • Open Access

    Public attitudes toward sharing health data for artificial intelligence: Differences by data type and sector in the Health in Central Denmark cohort

    by Jonas Rosborg Schaarup, Anders Aasted Isaksen, Adam Hulman

    Aims We aimed to examine public perceptions of sharing various types of health data relevant for AI development, including electronic health records, audio recordings of consultations, medical images, and genetic information, with actors from either the public or the private sectors. Methods We analysed data from 38,740 participants of the Health in Central Denmark survey conducted in 2024. Participants were asked whether they would share different types of health data with an AI solution in healthcare. Each participant was randomised to either of two versions of the scenario and question where the AI application was developed in the public or private sector. Descriptive results (proportions and percentages) were weighted to represent the background population of approx. 1 million people in the Central Denmark Region. The association between randomization group (data recipient) and data sharing attitude ('Yes', 'No', 'Don't know') was analysed using multinomial logistic regression with 'Don't know' as reference category. Results Participants were most willing to share medical images (46%), followed by text from patient journals (39%), genetic information (35%), and audio recordings (27%). There were 12-16% higher proportions of willingness to share with public institutions than with private institutions. A high level of uncertainty was observed for all data types (29-36%) regardless of data recipient. Odds ratios ranged from 1.37 to 1.78 for responding 'Yes', and from 0.51 to 0.67 for responding 'No' to sharing data with public institutions compared to private institutions. Conclusions Public acceptance of health data sharing for AI depends on both the perceived sensitivity of the data and the institutional context of use. Strong public governance, transparent safeguards, and clear communication about data use may be important for maintaining trust and enabling responsible development of AI in healthcare.

  • Open Access

    Trust in large language model-based solutions in healthcare among people with and without diabetes: a cross-sectional survey from the Health in Central Denmark cohort

    by Jonas Rosborg Schaarup, Anders Aasted Isaksen, Kasper Norman, Lasse Bjerg, Adam Hulman

    Objective: Large language models have gained significant public awareness since ChatGPT’s release in 2022. This study describes the perception of chatbots in healthcare among people with and without diabetes. Methods and analysis: In 2024, an online survey was sent to 136 229 people, aged 18–89 years in the Health in Central Denmark cohort, including equestions related to the perception of artificial intelligence (AI) and chatbots. Questions assessed trust in chatbots in various healthcare scenarios (lifestyle, diagnostic, contact with general practitioner (GP) and emergency contact) alongside participants’ level of experience with ChatGPT. In one item, participants were randomly presented with either a more severe (emergency) or less severe (GP contact) scenario. We used multinomial logistic regression to investigate the association of diabetes status and demographic characteristics with trust in chatbots in different scenarios. Results: 39 109 participants completed the questionnaire. The majority were aware of AI (94%), although fewer had heard of ChatGPT (76%), and only 21% had tried it. Most participants trusted chatbots with involvement of healthcare professionals (HCPs) (49%–55%), while few trusted without them (3%–6%). The degree of trust depended on the severity of the scenario, demonstrated by lower odds (OR 0.63, 95% CI 0.60 to 0.66) of trusting the chatbot under the control of HCPs in emergency care compared with contact with the GP. Type 2 diabetes, but not type 1 diabetes, was associated with less trust in chatbots than people without diabetes. Moreover, age, sex, duration of education and experience with ChatGPT also had an impact on trust. Conclusion: Chatbots are seen as supportive tools among public users when controlled by HCPs but are met with more scepticism in more severe situations. Digital exclusion risks and demographic differences, such as age, sex and disease-specific conditions (eg, type 2 diabetes), need to be addressed to ensure equitable and meaningful implementation.

  • osdc: Open Source Diabetes Classifier for Danish Registers

    by Signe Kirk Brødbæk, Anders Aasted Isaksen, Luke William Johnston

    The algorithm first identifies a population of individuals from Danish register data with any type of diabetes as individuals with two or more inclusion events. Then, it splits this population into individuals with either type 1 diabetes or type 2 diabetes by identifying individuals with type 1 diabetes and classifying the remainder of the diabetes population as having type 2 diabetes.

  • Open Access

    Changes in public perception of artificial intelligence in healthcare after exposure to ChatGPT

    by Anders Aasted Isaksen, Jonas Rosborg Schaarup, Lasse Bjerg, Adam Hulman

    The public perception of artificial intelligence (AI) in healthcare is key to its large-scale acceptance and implementation. This study investigated how exposure to ChatGPT changed public perception of AI in healthcare, using baseline and follow-up data from 5899 survey participants reporting their perception of AI in 2022 (before ChatGPT’s launch) and 2024, and ChatGPT use in 2024. Multinomial multivariate logistic regression was used to model how exposure to ChatGPT use affected changes in perception of AI. At follow-up, 1195 individuals (20%) had been exposed to ChatGPT use, which was associated with higher odds of changing perception of AI to beneficial (OR 3.21 [95% CI: 2.34–4.40]) among individuals who were unsure at baseline, and lower odds of changing to uncertainty from more defined baseline perceptions. This study demonstrates the potential for reducing uncertainty and improving public perception of AI in healthcare through exposure to AI tools.


  • Open Access

    Trust in large language model-based solutions in healthcare among people with and without diabetes: a cross-sectional survey from the Health in Central Denmark cohort

    by Jonas Rosborg Schaarup, Anders Aasted Isaksen, Kasper Norman, Lasse Bjerg, Adam Hulman

    Background: Large language models have gained significant public awareness since ChatGPT’s release in 2022. This study describes the perception of chatbot-assisted healthcare among people with and without diabetes. Methods: In 2024, an online survey was sent to 136,229 people, aged 18–89 years in the Health in Central Denmark cohort, including eight questions related to the perception of AI and chatbots. Questions assessed trust in chatbots in various healthcare scenarios (lifestyle, diagnostic, contact with general practitioner (GP), and emergency contact) alongside participants’ level of experience with ChatGPT. In one item, participants were randomly presented with either a more severe (emergency) or less severe (GP contact) scenario. We used multinomial logistic regression to investigate the association of diabetes status and demographic characteristics with trust in chatbots in different scenarios. Findings: 39,109 participants completed the questionnaire. The majority were aware of AI (94%), though fewer had heard of ChatGPT (76%), and only 21% had tried it. Most participants trusted chatbots with involvement of healthcare professionals (HCP) (49-55%), while few trusted without them (3–6%). The degree of trust depended on the severity of the scenario, demonstrated by lower odds (OR: 0.63 [95% CI: 0.60: 0.66]) of trusting the chatbot under the control of HCP in emergency care compared to contact with the general practitioner. Type 2 diabetes but not type 1 diabetes was associated with less trust in chatbots than people without diabetes. Moreover, age, sex, education, and experience with ChatGPT also had an impact on trust. Interpretation: Chatbots are seen as supportive tools among public users when controlled by HCPs but are met with more skepticism in more severe situations. Digital exclusion risks and demographic differences, such as age, sex, and disease-specific conditions (e.g., type 2 diabetes) needs, must be addressed to ensure equitable and meaningful implementation. Research in Context: Evidence before this study: Earlier studies have highlighted the generally positive attitudes of patients and the public towards the applications of artificial intelligence (AI) in healthcare. However, it noted a lack of clear characteristics associated with the acceptance of AI, with many patients preferring AI solutions to remain under human supervision rather than fully replacing healthcare professionals (HCPs). Since ChatGPT emerged in 2022, AI tools have been widely available to the general public, and many healthcare-specific chatbots are now being evaluated in random control trails. Some patients are already turning to tools like ChatGPT for medical advice, further underscoring the need to understand user perceptions, particularly in relation to diabetes and other characteristics, as these technologies may become integrated into care. Our earlier study showed that among AI applications, chatbots were the most controversial when used in emergency care. Thus, understanding the perception of chatbots in different healthcare contexts is needed, as the level of controversy may depend on their specific role in healthcare. Added value of this study: Our study expands on previous work by engaging a larger cohort of 39,109 participants, which includes a comprehensive representation of older adults and individuals with and without diabetes. Our survey was conducted between February-May 2024, a time when ChatGPT had been accessible for more than 1 year. We assessed trust in chatbot-based healthcare solutions, revealing that, while the majority accepted chatbot assistance under human control, individuals with type 2 diabetes exhibited less trust in such applications compared to those without diabetes or type 1 diabetes. Our findings underscore that the severity and acuteness of healthcare scenarios influenced trust levels. Implications of all available evidence: Our findings suggest that while AI and chatbots are becoming widely available, uncertainty about their benefits and risks in healthcare persists. People view healthcare professionals as playing an important role in supporting them, particularly in severe scenarios, toward adopting chatbot solutions. A patient-centered approach is necessary, with tailored solutions to address variations in trust based on factors such as diabetes status, age, sex, and education. Ensuring the involvement of vulnerable populations, such as the elderly and those with type 2 diabetes, is key to avoiding digital exclusion and making chatbot solutions accessible and meaningful.


  • Open Access

    Changes in public perception of AI in healthcare after exposure to ChatGPT

    by Anders Aasted Isaksen, Jonas Rosborg Schaarup, Lasse Bjerg, Adam Hulman

    Background Artificial intelligence (AI) is expected to become an integral part of healthcare services, and the widespread adoption of AI tools in all areas of life is making AI accessible to the general public. Public perception of the benefits and risks of AI in healthcare is key to large-scale acceptance and implementation, and is increasingly influenced by first-hand experiences of AI. The aim of this study was to assess how exposure to ChatGPT changed public perception of AI in healthcare. Methods We used baseline and follow-up data from 5,899 survey participants, who reported their perception of AI in 2022 and 2024, and ChatGPT use in 2024. Administrative and healthcare data from nationwide Danish registers was used for weighting and adjustment. Multinomial multivariate logistic regression was used to model how exposure to ChatGPT use affected changes in perception of AI. Results At baseline (before ChatGPT’s launch) 2,236 individuals (37%) were unsure of the benefits and risks of AI in healthcare, 2,384 (40%) perceived net benefits, 1,083 (18%) perceived benefits and risks as equal, and 196 (3.3%) perceived net risks. At follow-up, 1,195 individuals (20%) had been exposed to ChatGPT use, which was associated with higher odds of changing perception of AI to benefits (OR 3.21 [95% CI: 2.34-4.40]) among individuals who were unsure at baseline, and lower odds of changing to uncertainty from more defined baseline perceptions (from benefits (OR 0.32 [0.24-0.42]), equal (OR 0.47 [0.32-0.69]) and risks (OR 0.27 [0.08-0.98])). Conclusion Exposure to ChatGPT was associated with a change towards positive perception of benefits and risks of AI in healthcare among individuals who were uncertain prior to exposure, and individuals with more defined perceptions of AI were less likely to become uncertain after exposure to ChatGPT.