Management approaches are needed to prepare intervention data sets for research. We identified four management approaches and applied them to Omaha System intervention data from 15 home care agencies (621,385 interventions provided to 2,862 patients). Classifying intervention data created differing numbers of distinct groups for deductive approaches labeled as action category (four groups), theoretical (5), and clinical expert consensus (23). One inductive, data-driven approach generated 150 groups of interventions, of which 24 were meaningful and unique. Interventions in deductive groups were mutually exclusive, and approaches mapped readily according to intervention action terms. The novel, overlapping, inductive groups consisted of diverse actions for multiple problems. The four management approaches created meaningful intervention groups to be employed in future outcomes evaluation studies.

The large number of available MRI sequences means patients cannot realistically undergo them all, so the range of sequences to be acquired during a scan are protocolled based on clinical details. Adapting this to unexpected findings identified early on in the scan requires experience and vigilance. We investigated whether deep learning of the images acquired in the first few minutes of a scan could provide an automated early alert of abnormal features. Anatomy sequences from 375 CMR scans were used as a training set. From these, we annotated 1500 individual slices and used these to train a convolutional neural network to perform automatic segmentation of the cardiac chambers, great vessels and any pleural effusions. 200 scans were used as a testing set. The system then assembled a 3D model of the thorax from which it made clinical measurements to identify important abnormalities. The system was successful in segmenting the anatomy slices (Dice 0.910) and identified multiple features which may guide further image acquisition. Diagnostic accuracy was 90.5% and 85.5% for left and right ventricular dilatation, 85% for left ventricular hypertrophy and 94.4% for ascending aorta dilatation. The area under ROC curve for diagnosing pleural effusions was 0.91. We present proof-of-concept that a neural network can segment and derive accurate clinical measurements from a 3D model of the thorax made from transaxial anatomy images acquired in the first few minutes of a scan. This early information could lead to dynamic adaptive scanning protocols, and by focusing scanner time appropriately and prioritizing cases for supervision and early reporting, improve patient experience and efficiency.

Introduction Adrenal Insufficiency (AI), especially iatrogenic-AI, is a treatable cause of mortality. The difficulty in obtaining 9am cortisol levels means samples are taken at suboptimal times, including a substantial proportion in the afternoon. Low afternoon cortisol levels often provoke short Synacthen Tests (SSTs). It is important that this does not lead to patients misdiagnosed with AI, exposing them to the excess mortality and morbidity of inappropriate steroid replacement therapy. Methods This retrospective study collected 60,178 cortisol results. Medical records, including subsequent SSTs of initial cortisol results measured after midday were reviewed. Results ROC analysis (AUC- 0.89) on 6531 suitable cortisol values showed that a limit of <201.5nmol/L achieved a sensitivity and specificity of 95.6% and 72.6%, whilst a limit of <234nmol/L had a sensitivity of 100% and a specificity of 59.5%. Out of 670 SSTs, 628 patients passed. Of these, 140 would have otherwise failed if only their 30-minute cortisol was assessed without the 60-minute value. A 30-minute and 60-minute SST cortisol cut-off of 366.5nmol/L and 418.5nmol/L respectively, can achieve a sensitivity of >95% on the Abbott analyser platform. Conclusion An afternoon cortisol >234nmol/L excludes AI on Abbott analyser platforms. In patients who have an afternoon cortisol <234nmol/L, including both a 30-minute and a 60-minute SST cortisol values prevents unnecessary glucocorticoid replacement therapy in 22.3% of individuals in this study. The Abbott analyser SST cortisol cut-offs used to define AI should be 366.5nmol/L and 418.5nmol/L at 30- and 60-minutes respectively. All patients remained well subsequently with at least one year longitudinal follow up.

eating and body image were rarely discussed. 118 referrals were made as a result of HEADSS assessments; which constituted 54% of all encounters in which at least 1 category was screened. Referrals were limited in their scope, the majority being made to Social Services and safeguarding; smaller numbers were made to CAMHS, Red Thread and local Sexual Health services.