ObjectiveCurrent Endocrine Society Clinical Practice Guidelines use a specific aldosterone/renin ratio (ARR) threshold to screen for primary aldosteronism (a treatable disease causing up to 15% of hypertension in primary care) in all patients. We sought to characterize demographic variations in the ARR, hypothesizing a need for age‐ and sex‐specific reference ranges to improve the accuracy of the test.DesignRetrospective cross‐sectional analysis of ARR measurements at a single tertiary hospital from December 2016 to June 2018.PatientsA total of 442 patients with clinically indicated ARR were included, after excluding those who were on spironolactone or the oral contraceptive pill, were pregnant or had an existing adrenal condition.MeasurementsAldosterone, renin and the ARR.ResultsAmong those aged 20‐39 years (n = 74), females had significantly higher median aldosterone (369 vs 244 pmol/L, P = .028), lower median renin (17.0 vs 27.6 mIU/L, P = .034) and higher median ARR (20.7 vs 10.3 (pmol/L)/(mIU/L), P = .001) than males, despite having lower systolic (135 vs 145 mmHg, P = .021) and diastolic (89 vs 96.5 mmHg, P = .007) blood pressure. The ≥ 60‐year age group (n = 157) also had significant sex differences in the ARR. With increasing age (20‐39 vs ≥ 60 years), there was a significant fall in plasma aldosterone in females (369 pmol/L vs 264 pmol/L, P = .005), with no change observed in males.ConclusionsFor those 20‐39 years old, aldosterone and the ARR are significantly higher in females despite a lower systolic and diastolic BP, highlighting the potential for false‐positive results. Our findings indicate the need for prospective studies with a control population to define age‐ and sex‐specific ARR reference ranges.

Artificial intelligence (AI) offers much promise for improving healthcare. However, it runs the looming risk of causing individual and societal harms; for instance, exacerbating inequalities amongst minority groups, or enabling compromises in the confidentiality of patients’ sensitive data. As such, there is an expanding, unmet need for ensuring AI for healthcare is developed in concordance with human values and ethics. Augmenting “principle-based” guidance that highlight adherence to ethical ideals (without necessarily offering translation into actionable practices), we offer a solution-based framework for operationalising ethics in AI for healthcare. Our framework is built from a scoping review of existing solutions of ethical AI guidelines, frameworks and technical solutions to address human values such as self-direction in healthcare. Our view spans the entire length of the AI lifecycle: data management, model development, deployment and monitoring. Our focus in this paper is to collate actionable solutions (whether technical or non-technical in nature), which can be steps that enable and empower developers in their daily practice to ensuring ethical practices in the broader picture. Our framework is intended to be adopted by AI developers, with recommendations that are accessible and driven by the existing literature. We endorse the recognised need for ‘ethical AI checklists’ co-designed with health AI practitioners, which could further operationalise the technical solutions we have collated. Since the risks to health and wellbeing are so large, we believe a proactive approach is necessary for ensuring human values and ethics are appropriately respected in AI for healthcare.

No abstract

No abstract

ObjectiveElectronic health records (EHRs) can improve patient care by enabling systematic identification of patients for targeted decision support. But, this requires scalable learning of computable phenotypes. To this end, we developed the feature engineering automation tool (FEAT) and assessed it in targeting screening for the under-diagnosed, under-treated disease primary aldosteronism.Materials and MethodsWe selected 1199 subjects receiving longitudinal care in one health system between 2007 and 2017 and classified them for hypertension (N=608), hypertension with unexplained hypokalemia (N=172), and apparent treatment-resistant hypertension (N=176) by chart review. We derived 331 features from EHR encounters, diagnoses, laboratories, medications, vitals, and notes. We modified FEAT to encourage model parsimony and compared its models’ performance and interpretability to that of expert-curated heuristics and conventional machine learning.ResultsFEAT models trained to replicate expert-curated heuristics had higher AUPRC scores than all other models (p < 0.001) except random forests and were smaller than all other models (p < 1e-6) except decision trees. FEAT models trained to predict chart review phenotypes exhibited similar AUPRC scores to penalized logistic regression while being substantially simpler than all other models (p < 1e-6). For treatment-resistant hypertension, FEAT learned a six-feature, clinically intuitive model that demonstrated an adjusted PPV of 0.73 and sensitivity of 0.54 in testing.DiscussionFEAT learns computable phenotypes that approach the performance of expert-curated heuristics and conventional machine learning without sacrificing interpretability.ConclusionBy constructing accurate and interpretable computable phenotypes at scale, FEAT has the potential to facilitate widespread, systematic clinical decision support.

ObjectivesPrimary aldosteronism (PA), the most common endocrine cause of hypertension, is associated with a higher risk of cardiovascular disease (CVD) than blood pressure (BP)-matched essential hypertension (EH). We aimed to compare the calculated risks of CVD in patients who had hypertension with PA or EH using CVD risk calculators, hypothesising that they will fail to recognise the increased CVD risk in PA.DesignCross-sectional analysis.SettingAn endocrine hypertension service in Victoria, Australia.ParticipantsPatients who had hypertension without CVD referred for the investigation of hypertension.Outcome measuresCalculated 5-year or 10-year CVD risk as predicted by the National Vascular Disease Prevention Alliance (NVDPA) algorithm, Framingham Risk Score, Pooled Cohort Equations and QRISK3.ResultsThose with PA (n=128) and EH (n=133), did not differ significantly in their calculated CVD risks with the NVDPA algorithm (moderate-to-high 5-year risk 36/100 vs 45/99, p=0.17); the Framingham Risk Score (median 10-year risk 7.72% (4.43%–12.95%) vs 6.84% (3.85%–10.50%), p=0.14); the Pooled Cohort Equations (median 10-year risk 9.45% (4.36%–15.37%) vs 7.90% (2.09%–14.73%), p=0.07); and QRISK3 (median 10-year risk 11.31% (7.22%–20.29%) vs 12.47% (5.10%–19.93%), p=0.51). Similarities persisted on regression analyses accounting for systolic BP.ConclusionsCVD risk algorithms do not reflect the increased risk of CVD in patients with PA, and likely underestimate the true risk of CVD among those with PA. Screening for PA, in addition to using the CVD risk algorithm in patients who had hypertension, may facilitate the targeted treatment of PA and minimisation of cardiovascular risk in affected individuals.

No abstract