Accurate diagnosis of bipolar disorder (BPD) is difficult in clinical practice, with an average delay between symptom onset and diagnosis of about 7 years. A depressive episode often precedes the first manic episode, making it difficult to distinguish BPD from unipolar major depressive disorder (MDD).

We use genome-wide association analyses (GWAS) to identify differential genetic factors and to develop predictors based on polygenic risk scores (PRS) that may aid early differential diagnosis.

Based on individual genotypes from case–control cohorts of BPD and MDD shared through the Psychiatric Genomics Consortium, we compile case–case–control cohorts, applying a careful quality control procedure. In a resulting cohort of 51 149 individuals (15 532 BPD patients, 12 920 MDD patients and 22 697 controls), we perform a variety of GWAS and PRS analyses.

Although our GWAS is not well powered to identify genome-wide significant loci, we find significant chip heritability and demonstrate the ability of the resulting PRS to distinguish BPD from MDD, including BPD cases with depressive onset (BPD-D). We replicate our PRS findings in an independent Danish cohort (iPSYCH 2015, N = 25 966). We observe strong genetic correlation between our case–case GWAS and that of case–control BPD.

We find that MDD and BPD, including BPD-D are genetically distinct. Our findings support that controls, MDD and BPD patients primarily lie on a continuum of genetic risk. Future studies with larger and richer samples will likely yield a better understanding of these findings and enable the development of better genetic predictors distinguishing BPD and, importantly, BPD-D from MDD.

Loneliness has become a major public health issue of the recent decades due to its severe impact on health and mortality. Little is known about the relation between loneliness and social anxiety. This study aimed (1) to explore levels of loneliness and social anxiety in the general population, and (2) to assess whether and how loneliness affects symptoms of social anxiety and vice versa over a period of five years.

The study combined data from the baseline assessment and the five-year follow-up of the population-based Gutenberg Health Study. Data of N = 15 010 participants at baseline (Mage = 55.01, s.d.age = 11.10) were analyzed. Multiple regression analyses with loneliness and symptoms of social anxiety at follow-up including sociodemographic, physical illnesses, and mental health indicators at baseline were used to test relevant covariates. Effects of loneliness on symptoms of social anxiety over five years and vice versa were analyzed by autoregressive cross-lagged structural equation models.

At baseline, 1076 participants (7.41%) showed symptoms of social anxiety and 1537 (10.48%) participants reported feelings of loneliness. Controlling for relevant covariates, symptoms of social anxiety had a small significant effect on loneliness five years later (standardized estimate of 0.164, p < 0.001). Vice versa, there was no significant effect of loneliness on symptoms of social anxiety taking relevant covariates into account.

Findings provided evidence that symptoms of social anxiety are predictive for loneliness. Thus, prevention and intervention efforts for loneliness need to address symptoms of social anxiety.

The Personalized Advantage Index (PAI) shows promise as a method for identifying the most effective treatment for individual patients. Previous studies have demonstrated its utility in retrospective evaluations across various settings. In this study, we explored the effect of different methodological choices in predictive modelling underlying the PAI.

Our approach involved a two-step procedure. First, we conducted a review of prior studies utilizing the PAI, evaluating each study using the Prediction model study Risk Of Bias Assessment Tool (PROBAST). We specifically assessed whether the studies adhered to two standards of predictive modeling: refraining from using leave-one-out cross-validation (LOO CV) and preventing data leakage. Second, we examined the impact of deviating from these methodological standards in real data. We employed both a traditional approach violating these standards and an advanced approach implementing them in two large-scale datasets, PANIC-net (n = 261) and Protect-AD (n = 614).

The PROBAST-rating revealed a substantial risk of bias across studies, primarily due to inappropriate methodological choices. Most studies did not adhere to the examined prediction modeling standards, employing LOO CV and allowing data leakage. The comparison between the traditional and advanced approach revealed that ignoring these standards could systematically overestimate the utility of the PAI.

Our study cautions that violating standards in predictive modeling may strongly influence the evaluation of the PAI's utility, possibly leading to false positive results. To support an unbiased evaluation, crucial for potential clinical application, we provide a low-bias, openly accessible, and meticulously annotated script implementing the PAI.

For more than 2 years, coronavirus disease (COVID-19) has forced worldwide health care systems to adapt their daily practice. These adaptations add to the already stressful demands of providing timely medical care in an overcrowded health care system. Specifically, the COVID-19 pandemic added stress to an already overwhelmed emergency and critical care health care workers (HCWs) on the front lines during the first wave of the pandemic.

This study assessed comparative subjective and objective stress among frontline HCWs using a visual analog scale and biometric data, specifically heart rate variability (HRV).

This is a prospective, observational study using surveys and heart rate monitoring among HCWs who work in 3 frontline health care units (emergency department, mobile intensive care unit, and intensive care unit) in the University Hospital of Clermont-Ferrand, France. Two sessions were performed: 1 during the first wave of the pandemic (April 10 to May 10, 2020) and 1 after the first wave of the pandemic (June 10 to July 15, 2020).

The primary outcome is the difference in stress levels between the 2 time points. Secondary objectives were the impact of overcrowding, sociodemographics, and other variables on stress levels. We also assessed the correlation between subjective and objective stress levels.

Among 199 HCWs, 98 participated in biometric monitoring, 84 had biometric and survey data, and 12 with only biometric data. Subjective stress was higher during the second time point compared to the first (4.39 ± 2.11 vs 3.16 ± 2.34, P = 0.23). There were higher objective stress levels with a decrease in HRV between the first and the second time points. Furthermore, we found higher patient volumes as a source of stress during the second time point. We did not find any significant correlation between subjective and objective stress levels.

HCWs had higher stress levels between the 2 waves of the pandemic. Overcrowding in the emergency department is associated with higher stress levels. We did not find any correlation between subjective and objective stress among intensive care and emergency HCWs during the first wave of the pandemic.

Artificial intelligence (AI) is seen as one of the major disrupting forces in the future healthcare system. However, assessment of the value of these new technologies is still unclear and no agreed international HTA-based guideline exists. Therefore, a Model for ASsessing the value of AI (MAS-AI) in medical imaging was developed by a multidisciplinary group of experts and patient representatives.

The MAS-AI guideline is based on four steps. First a literature review of existing guides, evaluations, and assessments of the value of AI in the field of medical imaging (5,890 studies were assessed with 86 studies included in the scoping review). Next, interviews with leading researchers in AI in Denmark. The third step was two workshops where decision-makers, patient organizations and researchers discussed crucial topics when evaluating AI. Between workshops, the multidisciplinary team revised the model according to comments from workshop-participants. Last step is a validation workshop in Canada.

The MAS-AI guideline has three parts. There are two steps covering nine domains and then advises for the evaluation process. Step 1 contains a description of patients, how the AI-model was developed, and initial ethical and legal considerations. Finishing the four domains in Step 1 is a prerequisite for moving to step 2. In step 2, a multidisciplinary assessment of outcomes of the AI-application is done for the five remaining domains: safety, clinical aspects, economics, organizational aspects and patient aspects. The last part, is five advices to facilitate a good evaluation process.

We have developed an HTA based framework to support the prospective phase while introducing novel AI technologies into healthcare in medical imaging. MAS-AI can assist HTA organizations (and companies) in selecting the relevant domains and outcome measures in the assessment of AI applications. It is important to ensure uniform and valid decisions regarding the adoption of AI technology with a structured process and tool. MAS-AI can help support these decisions and provide greater transparency for all parties involved.

Palliative sedation (PS) is an intrusive measure to relieve patients at the end of their life from otherwise untreatable symptoms. Intensive discussion of the advantages and limitations of palliative care with the patients and their relatives should precede the initiation of PS since PS is terminated by the patient’s death in most cases. Drugs for PS are usually administered intravenously. Midazolam is widely used, either alone or in combination with other substances. PS can be conducted in both inpatient and outpatient settings; however, a quality analysis comparing both modalities was missing so far.

This prospective observational study collected data from patients undergoing PS inpatient at the palliative care unit (PCU, n = 26) or outpatient at a hospice (n = 2) or at home (specialized outpatient palliative care [SAPV], n = 31) between July 2017 and June 2018. Demographical data, indications for PS, and drug protocols were analyzed. The depth of sedation according to the Richmond Agitation Sedation Scale (RASS) and the degree of satisfaction of staff members and patient’s relatives were included as parameters for quality assessment.

Patients undergoing PS at the PCU were slightly younger compared to outpatients (hospice and SAPV combined). Most patients suffered from malignant diseases, and midazolam was the backbone of sedation for inpatients and outpatients. The median depth of sedation was between +1 and −3 according to the RASS with a trend to deeper sedation prior to death. The median degree of satisfaction was “good,” scored by staff members and by patient’s relatives. Significant differences between inpatients and outpatients were not seen in protocols, depth of sedation, and degree of satisfaction.

The data support the thesis that PS is possible for inpatients and outpatients with comparable results. For choosing the best place for PS, other aspects such as patient’s and relative’s wishes, stress, and medical reasons should be considered.

Artificial intelligence (AI) is seen as a major disrupting force in the future healthcare system. However, the assessment of the value of AI technologies is still unclear. Therefore, a multidisciplinary group of experts and patients developed a Model for ASsessing the value of AI (MAS-AI) in medical imaging. Medical imaging is chosen due to the maturity of AI in this area, ensuring a robust evidence-based model.

MAS-AI was developed in three phases. First, a literature review of existing guides, evaluations, and assessments of the value of AI in the field of medical imaging. Next, we interviewed leading researchers in AI in Denmark. The third phase consisted of two workshops where decision makers, patient organizations, and researchers discussed crucial topics for evaluating AI. The multidisciplinary team revised the model between workshops according to comments.

The MAS-AI guideline consists of two steps covering nine domains and five process factors supporting the assessment. Step 1 contains a description of patients, how the AI model was developed, and initial ethical and legal considerations. In step 2, a multidisciplinary assessment of outcomes of the AI application is done for the five remaining domains: safety, clinical aspects, economics, organizational aspects, and patient aspects.

We have developed an health technology assessment-based framework to support the introduction of AI technologies into healthcare in medical imaging. It is essential to ensure informed and valid decisions regarding the adoption of AI with a structured process and tool. MAS-AI can help support decision making and provide greater transparency for all parties.

Cognitive Bias Modification for paranoia (CBM-pa) is a novel, theory-driven psychological intervention targeting the biased interpretation of emotional ambiguity associated with paranoia. Study objectives were (i) test the intervention's feasibility, (ii) provide effect size estimates, (iii) assess dose–response and (iv) select primary outcomes for future trials.

In a double-blind randomised controlled trial, sixty-three outpatients with clinically significant paranoia were randomised to either CBM-pa or an active control (text reading) between April 2016 and September 2017. Patients received one 40 min session per week for 6 weeks. Assessments were given at baseline, after each interim session, post-treatment, and at 1- and 3-months post-treatment.

A total of 122 patients were screened and 63 were randomised. The recruitment rate was 51.2%, with few dropouts (four out of 63) and follow-up rates were 90.5% (1-month) and 93.7% (3-months). Each session took 30–40 min to complete. There was no statistical evidence of harmful effects of the intervention. Preliminary data were consistent with efficacy of CBM-pa over text-reading control: patients randomised to the intervention, compared to control patients, reported reduced interpretation bias (d = −0.48 to −0.76), improved symptoms of paranoia (d = −0.19 to −0.38), and lower depressed and anxious mood (d = −0.03 to −0.29). The intervention effect was evident after the third session.

CBM-pa is feasible for patients with paranoia. A fully powered randomised control trial is warranted.

Case-only longitudinal studies are common in psychiatry. Further, it is assumed that psychiatric ratings and questionnaire results of healthy controls stay stable over foreseeable time ranges. For cognitive tests, improvements over time are expected, but data for more than two administrations are scarce.

We comprehensively investigated the longitudinal course for trends over time in cognitive and symptom measurements for severe mental disorders. Assessments included the Trail Making Tests, verbal Digit Span tests, Global Assessment of Functioning, Inventory of Depressive Symptomatology, the Positive and Negative Syndrome Scale, and the Young Mania Rating Scale, among others.

Using the data of control individuals (n = 326) from the PsyCourse study who had up to four assessments over 18 months, we modelled the course using linear mixed models or logistic regression. The slopes or odds ratios were estimated and adjusted for age and gender. We also assessed the robustness of these results using a longitudinal non-parametric test in a sensitivity analysis.

Small effects were detected for most cognitive tests, indicating a performance improvement over time (P < 0.05). However, for most of the symptom rating scales and questionnaires, no effects were detected, in line with our initial hypothesis.

The slightly but consistently improved performance in the cognitive tests speaks of a test-unspecific positive trend, while psychiatric ratings and questionnaire results remain stable over the observed period. These detectable improvements need to be considered when interpreting longitudinal courses. We therefore recommend recruiting control participants if cognitive tests are administered.