Few artificial intelligence (AI) clinical decision support systems (CDSSs) are ever evaluated in practice. Although some signal of clinical effectiveness may be needed to justify AI deployment and testing, such data are typically unavailable in early-stage research. This conundrum is especially relevant in the intensive care unit (ICU), where conditions like sepsis and acute respiratory distress syndrome (ARDS) require high-stakes decisions. Our group developed the AI ventilator assistant (AVA), a novel AI CDSS for patients with sepsis ARDS receiving invasive mechanical ventilation. But the promising results of predictive performance estimates are not sufficient to assess AVA’s clinical safety and appropriateness prior to future evaluation and deployment. Therefore, we propose a Clinician Turing Test as a novel validation approach to determine whether clinicians can distinguish AVA-generated treatment recommendations from those enacted by real human clinicians. If AVA’s recommendations are consistently indistinguishable from those of real clinicians, thereby ‘passing’ this Turing test, this would provide a strong preclinical signal of safety and appropriateness.

This multisite, randomised, electronic, vignette-based Phase 1b study will use a Clinician Turing Test design. We aim to recruit 350 critical care clinicians, including physicians and advanced practice providers from six US hospitals. Participants will review nine clinical vignettes of patients with sepsis and ARDS derived from the Molecular Epidemiology of Severe Sepsis in the ICU cohort and an associated profile of a suggested treatment plan. For each participant–vignette combination, the source of the treatment profile will be randomly assigned (AI-generated by AVA vs the actually enacted treatment from real human clinicians) in a 1:1 allocation. The primary endpoint is the participants’ accuracy in identifying whether a treatment profile was AI-generated or human-generated, assessed using equivalence testing through a mixed-effects logistic regression model with random effects for participants and vignettes. Secondarily, a fitted binary classifier will assess discrimination ability using the C-statistic. Secondary endpoints include clinicians’ perceptions of the safety and appropriateness of the treatment profiles, confidence in distinguishing AI-generated and human-generated recommendations, interest in AI CDSSs for sepsis and ventilator management and the time to complete the survey. This novel Phase 1b design provides preliminary but essential information about an AI CDSS’s clinical appropriateness without the risk or cost of actual deployment, thereby informing decisions about future clinical implementation and evaluation in real clinical environments.

This protocol was approved by the Institutional Review Board of the University of Pennsylvania (Protocol #858201). Results are expected in 2026 and will be submitted for publication in peer-reviewed journals and presented at scientific conferences.

NCT07025096.

Osteogenesis imperfecta (OI) is the most common inherited cause of bone fragility (approximately 1 in 16 000). People with OI suffer bone fragility causing fractures, pain and deformity; sarcopenia causing fatigue and poor endurance; aortic root dilatation and hearing loss. No drug currently has market authorisation to treat OI in Europe. Current standard-of-care is multidisciplinary, with pharmacological interventions—primarily bisphosphonates—directed at increasing bone mass; however, such interventions are of equivocal efficacy. The structural damage that can accumulate as a result of repeated fractures over time may not be reversible. The lack of a treatment with clearly defined efficacy in terms of reducing fracture frequency or the sarcopenia, that is increasingly recognised in this condition, leads to the consideration of alternatives based on what is known about the molecular pathophysiology of the condition. For reasons that are currently unclear, transforming growth factor beta (TGFβ) pathway signalling is increased in OI, and both studies in mouse models and more recently also in humans suggest that reducing TGFβ pathway signalling could be of benefit in OI. This demonstrator project tests the hypothesis that losartan, an antihypertensive agent known to reduce circulating TGFβ, will reduce bone turnover and bone loss and have a positive effect on muscle function and quality of life in adults and older adolescents with OI.

This is a phase 2/pilot, open-label, dose-escalating study. This study aims to identify the effective dose for losartan in this population to inform the design of a pivotal phase III study. The study aims to recruit 30 adolescents and adults aged 16 years and above with OI across secondary care study sites in the UK and Italy. Participants will be recruited from the patient populations attending for treatment of OI at the participating hospital sites or referred by clinicians at the Participant Identification Centres (PIC sites). Participants will be randomised to one of three ‘final doses’—25, 50 or 75 mg losartan once daily. All participants will start on 25 mg once daily. Those assigned to higher ‘final doses’ will increase in 25 mg once daily increments on day 8 and day 15 following safety assessments. The primary outcome measures are to establish the effective dose of losartan in OI patients, based on maximal reduction in the bone resorption marker carboxy-terminal crosslink of type I collagen telopeptide (CTX) over the 24-week period of the study.

Secondary outcome measures are to determine the changes in proxy efficacy outcomes for bone (turnover, mass, architecture and strength) using blood tests, high-resolution peripheral quantitative CT (HRpQCT), dual-energy X-ray absorptiometry (DXA) and muscle (strength) using the ‘Timed Up and Go’ test. In addition, the changes in quality of life, including pain and fatigue, will be evaluated by using a disease-specific tool (OI-QOL) and a validated generic tool (EQ-5D-5L-VAS).

In the UK, the study protocol and amendments have been approved by the London Bridge Research Ethics Committee (REC reference: 23/LO/015) and by the Medicines and Healthcare products Regulatory Agency (MHRA). In Italy, the study protocol and amendments have been approved by the Italian and European ethics and regulatory authorities (Clinical Trials Information System European Union (CTIS EU) portal according to EU Regulation 536/2014). Final version of study protocol: Version 3.2, 05.03.2025. Final results will be disseminated in peer-reviewed journals through local OI, orthopaedic and other relevant clinical networks and at national and international meetings. Sheffield Children’s National Health Service Foundation Trust (UK) and Istituto di Ricovero e Cura a Carattere Scientifico (IRCCS) Istituto Ortopedico Rizzoli (Italy) are the joint study sponsors.

ISRCTN (ISRCTN13317811).

This study aimed to (1) assess Lumos data quality, a New South Wales (NSW) statewide linked health data asset; and (2) determine sociodemographic variation in health service utilisation of general practice, emergency department and admitted services.

A retrospective cohort study using Lumos, a linked health data asset.

A representative statewide sample population of NSW, Australia.

People residing within NSW with an electronic health record at a Lumos participating general practice between January 2010 and June 2023.

Data quality indicators of Lumos including completeness, representativeness against NSW population data, consistency and timeliness. Furthermore, variation in general practice visits, emergency department presentations and hospital admission rates stratified by age, sex, rurality and Index of Relative Socio-economic Disadvantage (IRSD)—a measure of socioeconomic status used in Australia, where lower values represent greater relative disadvantage across a range of metrics such as education and income.

At the time of analysis, Lumos included records from 5.2 million unique patients, representing half (49.7%) of the NSW resident population. Limiting data to 2022, the Lumos population distribution broadly aligned with the 2021 Census except for IRSD quintile four and five which were under-represented (15.0% vs 20.4% (standardised difference –0.14)), and over-represented (29.7% vs 19.9% (standardised difference 0.23)), respectively. Age and greater relative disadvantage were associated with higher rates of general practice visits and hospital admissions. Greater relative disadvantage was also associated with higher rates of emergency department presentations.

Lumos’s ability to overcome historical limitations of separately managed health data in Australia and its demonstrated data quality present an opportunity to enhance health system policy and planning in NSW. The variation in service utilisation across primary and tertiary care by population and geography apparent in Lumos reinforces the need for tailored service planning.