Impact of Transfer Learning Using Local Data on Performance of a Deep Learning Model for Screening Mammography.

"Just Accepted" papers have undergone full peer review and have been accepted for publication in Radiology: Artificial Intelligence. This article will undergo copyediting, layout, and proof review before it is published in its final version. Please note that during production of the final copyedited article, errors may be discovered which could affect the content. Purpose To investigate the issues of generalizability and replication of deep learning (DL) models by assessing performance of a screening mammography DL system developed at New York University (NYU) on a local Australian dataset. Materials and Methods In this retrospective study, all individuals with biopsy and surgical pathology-proven lesions and age-matched controls were identified from a South Australian public mammography screening program (January 2010 to December 2016). The primary outcome was DL system performance, measured with the area under the receiver operating characteristic curve (AUC), in classifying invasive breast cancer or ductal carcinoma in situ (n = 425) from no malignancy (n = 490) or benign lesions (n = 44) in age-matched controls. The NYU system, including models without (NYU1) and with (NYU2) heatmaps, was tested in its original form, after training from scratch (without transfer learning; TL), after retraining with TL. Results The local test set comprised 959 individuals (mean age, 62.5 years [SD, 8.5]; all female). The original AUCs for the NYU1 and NYU2 models were 0.83 (95%CI = 0.82-0.84) and 0.89 (95%CI = 0.88-0.89), respectively. When applied in their original form to the local test set, the AUCs were 0.76 (95%CI = 0.73-0.79) and 0.84 (95%CI = 0.82-0.87), respectively. After local training without TL, the AUCs were 0.66 (95%CI = 0.62-0.69) and 0.86 (95%CI = 0.84-0.88). After retraining with TL, the AUCs were 0.82 (95%CI = 0.80-0.85) and 0.86 (95%CI = 0.84-0.88). Conclusion A deep learning system developed using a U.S. dataset showed reduced performance when applied 'out of the box' to an Australian dataset. Local retraining with transfer learning using available model weights improved model performance. ©RSNA, 2024.

Prognostic modeling in early rheumatoid arthritis: reconsidering the predictive role of disease activity scores.

OBJECTIVE: In this prospective cohort study, we provide several prognostic models to predict functional status as measured by the modified Health Assessment Questionnaire (mHAQ). The early adoption of the treat-to-target strategy in this cohort offered a unique opportunity to identify predictive factors using longitudinal data across 20 years. METHODS: A cohort of 397 patients with early RA was used to develop statistical models to predict mHAQ score measured at baseline, 12 months, and 18 months post diagnosis, as well as serially measured mHAQ. Demographic data, clinical measures, autoantibodies, medication use, comorbid conditions, and baseline mHAQ were considered as predictors. RESULTS: The discriminative performance of models was comparable to previous work, with an area under the receiver operator curve ranging from 0.64 to 0.88. The most consistent predictive variable was baseline mHAQ. Patient-reported outcomes including early morning stiffness, tender joint count (TJC), fatigue, pain, and patient global assessment were positively predictive of a higher mHAQ at baseline and longitudinally, as was the physician global assessment and C-reactive protein. When considering future function, a higher TJC predicted persistent disability while a higher swollen joint count predicted functional improvements with treatment. CONCLUSION: In our study of mHAQ prediction in RA patients receiving treat-to-target therapy, patient-reported outcomes were most consistently predictive of function. Patients with high disease activity due predominantly to tenderness scores rather than swelling may benefit from less aggressive treatment escalation and an emphasis on non-pharmacological therapies, allowing for a more personalized approach to treatment. Key Points • Long-term use of the treat-to-target strategy in this patient cohort offers a unique opportunity to develop prognostic models for functional outcomes using extensive longitudinal data. • Patient reported outcomes were more consistent predictors of function than traditional prognostic markers. • Tender joint count and swollen joint count had discordant relationships with future function, adding weight to the possibility that disease activity may better guide treatment when the components are considered separately.

Atrial Fibrillation Health Literacy Questionnaire (AFHLQ): The development of an AF-specific health literacy questionnaire.

Background: Health literacy is a key enabler of effective behavioural modification in chronic diseases. While patient reported outcome measures (PROMs) exists for patient with atrial fibrillation (AF), none address risk factors comprehensively. The aim of the study was to develop and qualitatively validate a disease specific PROM that incorporates knowledge on risk factors and assesses interactive and critical health literacy of people living with AF. Methods: The 47-item Atrial Fibrillation Health Literacy Questionnaire (AFHLQ) was developed and validated through a qualitative research design. Expert and Consumer focus groups, each consisting of seven participants provided opinion. Results: The 47-item questionnaire consists of 5 domains: (1) what is AF, (2) what are the symptoms of AF, (3) why do people get AF, (4) management of AF, and (5) what measures can slow or prevent the progression of AF. Recommendations resulted in several changes to the original 47 item list during the qualitative validation process: 13 original items were removed, and 13 new items were added. The response categories were also simplified from a Likert scale to "yes", "no" or "don't know". Conclusion: A 47-item AFHLQ instrument was developed and validated with modifications made through clinical expert and consumer opinion. This tool has a potential to be used to evaluate and guide interventions at a clinical and population level to understand and improve AF health literacy and outcomes.

Machine learning and deep learning predictive models for long-term prognosis in patients with chronic obstructive pulmonary disease: a systematic review and meta-analysis.

BACKGROUND: Machine learning and deep learning models have been increasingly used to predict long-term disease progression in patients with chronic obstructive pulmonary disease (COPD). We aimed to summarise the performance of such prognostic models for COPD, compare their relative performances, and identify key research gaps. METHODS: We conducted a systematic review and meta-analysis to compare the performance of machine learning and deep learning prognostic models and identify pathways for future research. We searched PubMed, Embase, the Cochrane Library, ProQuest, Scopus, and Web of Science from database inception to April 6, 2023, for studies in English using machine learning or deep learning to predict patient outcomes at least 6 months after initial clinical presentation in those with COPD. We included studies comprising human adults aged 18-90 years and allowed for any input modalities. We reported area under the receiver operator characteristic curve (AUC) with 95% CI for predictions of mortality, exacerbation, and decline in forced expiratory volume in 1 s (FEV1). We reported the degree of interstudy heterogeneity using Cochran's Q test (significant heterogeneity was defined as p≤0·10 or I2>50%). Reporting quality was assessed using the TRIPOD checklist and a risk-of-bias assessment was done using the PROBAST checklist. This study was registered with PROSPERO (CRD42022323052). FINDINGS: We identified 3620 studies in the initial search. 18 studies were eligible, and, of these, 12 used conventional machine learning and six used deep learning models. Seven models analysed exacerbation risk, with only six reporting AUC and 95% CI on internal validation datasets (pooled AUC 0·77 [95% CI 0·69-0·85]) and there was significant heterogeneity (I2 97%, p<0·0001). 11 models analysed mortality risk, with only six reporting AUC and 95% CI on internal validation datasets (pooled AUC 0·77 [95% CI 0·74-0·80]) with significant degrees of heterogeneity (I2 60%, p=0·027). Two studies assessed decline in lung function and were unable to be pooled. Machine learning and deep learning models did not show significant improvement over pre-existing disease severity scores in predicting exacerbations (p=0·24). Three studies directly compared machine learning models against pre-existing severity scores for predicting mortality and pooled performance did not differ (p=0·57). Of the five studies that performed external validation, performance was worse than or equal to regression models. Incorrect handling of missing data, not reporting model uncertainty, and use of datasets that were too small relative to the number of predictive features included provided the largest risks of bias. INTERPRETATION: There is limited evidence that conventional machine learning and deep learning prognostic models demonstrate superior performance to pre-existing disease severity scores. More rigorous adherence to reporting guidelines would reduce the risk of bias in future studies and aid study reproducibility. FUNDING: None.

Leveraging pleiotropy to discover and interpret GWAS results for sleep-associated traits.

Genetic association studies of many heritable traits resulting from physiological testing often have modest sample sizes due to the cost and burden of the required phenotyping. This reduces statistical power and limits discovery of multiple genetic associations. We present a strategy to leverage pleiotropy between traits to both discover new loci and to provide mechanistic hypotheses of the underlying pathophysiology. Specifically, we combine a colocalization test with a locus-level test of pleiotropy. In simulations, we show that this approach is highly selective for identifying true pleiotropy driven by the same causative variant, thereby improves the chance to replicate the associations in underpowered validation cohorts and leads to higher interpretability. Here, as an exemplar, we use Obstructive Sleep Apnea (OSA), a common disorder diagnosed using overnight multi-channel physiological testing. We leverage pleiotropy with relevant cellular and cardio-metabolic phenotypes and gene expression traits to map new risk loci in an underpowered OSA GWAS. We identify several pleiotropic loci harboring suggestive associations to OSA and genome-wide significant associations to other traits, and show that their OSA association replicates in independent cohorts of diverse ancestries. By investigating pleiotropic loci, our strategy allows proposing new hypotheses about OSA pathobiology across many physiological layers. For example, we identify and replicate the pleiotropy across the plateletcrit, OSA and an eQTL of DNA primase subunit 1 (PRIM1) in immune cells. We find suggestive links between OSA, a measure of lung function (FEV1/FVC), and an eQTL of matrix metallopeptidase 15 (MMP15) in lung tissue. We also link a previously known genome-wide significant peak for OSA in the hexokinase 1 (HK1) locus to hematocrit and other red blood cell related traits. Thus, the analysis of pleiotropic associations has the potential to assemble diverse phenotypes into a chain of mechanistic hypotheses that provide insight into the pathogenesis of complex human diseases.

Artificial intelligence and the future of radiographic scoring in rheumatoid arthritis: a viewpoint.

Rheumatoid arthritis is an autoimmune condition that predominantly affects the synovial joints, causing joint destruction, pain, and disability. Historically, the standard for measuring the long-term efficacy of disease-modifying antirheumatic drugs has been the assessment of plain radiographs with scoring techniques that quantify joint damage. However, with significant improvements in therapy, current radiographic scoring systems may no longer be fit for purpose for the milder spectrum of disease seen today. We argue that artificial intelligence is an apt solution to further improve upon radiographic scoring, as it can readily learn to recognize subtle patterns in imaging data to not only improve efficiency, but can also increase the sensitivity to variation in mild disease. Current work in the area demonstrates the feasibility of automating scoring but is yet to take full advantage of the strengths of artificial intelligence. By fully leveraging the power of artificial intelligence, faster and more sensitive scoring could enable the ongoing development of effective treatments for patients with rheumatoid arthritis.

Pre-thrombectomy prognostic prediction of large-vessel ischemic stroke using machine learning: A systematic review and meta-analysis.

Introduction: Machine learning (ML) methods are being increasingly applied to prognostic prediction for stroke patients with large vessel occlusion (LVO) treated with endovascular thrombectomy. This systematic review aims to summarize ML-based pre-thrombectomy prognostic models for LVO stroke and identify key research gaps. Methods: Literature searches were performed in Embase, PubMed, Web of Science, and Scopus. Meta-analyses of the area under the receiver operating characteristic curves (AUCs) of ML models were conducted to synthesize model performance. Results: Sixteen studies describing 19 models were eligible. The predicted outcomes include functional outcome at 90 days, successful reperfusion, and hemorrhagic transformation. Functional outcome was analyzed by 10 conventional ML models (pooled AUC=0.81, 95% confidence interval [CI]: 0.77-0.85, AUC range: 0.68-0.93) and four deep learning (DL) models (pooled AUC=0.75, 95% CI: 0.70-0.81, AUC range: 0.71-0.81). Successful reperfusion was analyzed by three conventional ML models (pooled AUC=0.72, 95% CI: 0.56-0.88, AUC range: 0.55-0.88) and one DL model (AUC=0.65, 95% CI: 0.62-0.68). Conclusions: Conventional ML and DL models have shown variable performance in predicting post-treatment outcomes of LVO without generally demonstrating superiority compared to existing prognostic scores. Most models were developed using small datasets, lacked solid external validation, and at high risk of potential bias. There is considerable scope to improve study design and model performance. The application of ML and DL methods to improve the prediction of prognosis in LVO stroke, while promising, remains nascent. Systematic review registration: https://www.crd.york.ac.uk/prospero/display_record.php?ID=CRD42021266524, identifier CRD42021266524.

Targeted Genome Sequencing Identifies Multiple Rare Variants in Caveolin-1 Associated with Obstructive Sleep Apnea.

Rationale: Obstructive sleep apnea (OSA) is a common disorder associated with increased risk for cardiovascular disease, diabetes, and premature mortality. There is strong clinical and epidemiologic evidence supporting the importance of genetic factors influencing OSA but limited data implicating specific genes. Objectives: To search for rare variants contributing to OSA severity. Methods: Leveraging high-depth genomic sequencing data from the NHLBI Trans-Omics for Precision Medicine (TOPMed) program and imputed genotype data from multiple population-based studies, we performed linkage analysis in the CFS (Cleveland Family Study), followed by multistage gene-based association analyses in independent cohorts for apnea-hypopnea index (AHI) in a total of 7,708 individuals of European ancestry. Measurements and Main Results: Linkage analysis in the CFS identified a suggestive linkage peak on chromosome 7q31 (LOD = 2.31). Gene-based analysis identified 21 noncoding rare variants in CAV1 (Caveolin-1) associated with lower AHI after accounting for multiple comparisons (P = 7.4 × 10-8). These noncoding variants together significantly contributed to the linkage evidence (P < 10-3). Follow-up analysis revealed significant associations between these variants and increased CAV1 expression, and increased CAV1 expression in peripheral monocytes was associated with lower AHI (P = 0.024) and higher minimum overnight oxygen saturation (P = 0.007). Conclusions: Rare variants in CAV1, a membrane-scaffolding protein essential in multiple cellular and metabolic functions, are associated with higher CAV1 gene expression and lower OSA severity, suggesting a novel target for modulating OSA severity.

The clinical and genetic features of hereditary pancreatitis in South Australia.

OBJECTIVE: To characterise the clinical phenotypes and genetic variants of hereditary pancreatitis in people diagnosed in South Australia. DESIGN, SETTING, PARTICIPANTS: Cross-sectional study of people who received molecular diagnoses of hereditary pancreatitis from one of four major diagnostic services in South Australia, 1 January 2006 - 30 June 2021. MAIN OUTCOME MEASURES: Genotypic and clinical features of people with hereditary pancreatitis, including age at onset, attack frequency, pain indices, use of opioid medications, and physical and mental health impact of hereditary pancreatitis. RESULTS: We identified 44 people from ten families who received molecular diagnoses of hereditary pancreatitis during 2006-21 (including 25 Indigenous people [57%] and 27 women [61%]): 36 with PRSS1, five with SPINK1, and three with PRSS1 and SPINK1 mutations (determined by whole exome sequencing). Symptom onset before the age of ten years was reported by 37 people (84%). Pancreatitis-related pain during the preceding four weeks was described as moderate or high by 35 people (79%); 38 people regularly used opioids (86%). Fifteen patients had diabetes mellitus (34%), and eight had undergone pancreatic surgery (18%). The estimated prevalence of hereditary pancreatitis was 1.1 (95% CI, 0.72-1.4) cases per 100 000 population for non-Indigenous and 71 (95% CI, 66-77) cases per 100 000 population for Indigenous South Australians. Among people with adult-onset chronic pancreatitis admitted to South Australian public hospitals during 2001-2019, the proportions of Indigenous people (12%) and women (38%) were smaller than we report for hereditary pancreatitis. CONCLUSION: The estimated prevalence of hereditary pancreatitis in South Australia is higher than in Europe. PRSS1 gene mutations are important causes, particularly among Indigenous young people.

AI recognition of patient race in medical imaging: a modelling study.

BACKGROUND: Previous studies in medical imaging have shown disparate abilities of artificial intelligence (AI) to detect a person's race, yet there is no known correlation for race on medical imaging that would be obvious to human experts when interpreting the images. We aimed to conduct a comprehensive evaluation of the ability of AI to recognise a patient's racial identity from medical images. METHODS: Using private (Emory CXR, Emory Chest CT, Emory Cervical Spine, and Emory Mammogram) and public (MIMIC-CXR, CheXpert, National Lung Cancer Screening Trial, RSNA Pulmonary Embolism CT, and Digital Hand Atlas) datasets, we evaluated, first, performance quantification of deep learning models in detecting race from medical images, including the ability of these models to generalise to external environments and across multiple imaging modalities. Second, we assessed possible confounding of anatomic and phenotypic population features by assessing the ability of these hypothesised confounders to detect race in isolation using regression models, and by re-evaluating the deep learning models by testing them on datasets stratified by these hypothesised confounding variables. Last, by exploring the effect of image corruptions on model performance, we investigated the underlying mechanism by which AI models can recognise race. FINDINGS: In our study, we show that standard AI deep learning models can be trained to predict race from medical images with high performance across multiple imaging modalities, which was sustained under external validation conditions (x-ray imaging [area under the receiver operating characteristics curve (AUC) range 0·91-0·99], CT chest imaging [0·87-0·96], and mammography [0·81]). We also showed that this detection is not due to proxies or imaging-related surrogate covariates for race (eg, performance of possible confounders: body-mass index [AUC 0·55], disease distribution [0·61], and breast density [0·61]). Finally, we provide evidence to show that the ability of AI deep learning models persisted over all anatomical regions and frequency spectrums of the images, suggesting the efforts to control this behaviour when it is undesirable will be challenging and demand further study. INTERPRETATION: The results from our study emphasise that the ability of AI deep learning models to predict self-reported race is itself not the issue of importance. However, our finding that AI can accurately predict self-reported race, even from corrupted, cropped, and noised medical images, often when clinical experts cannot, creates an enormous risk for all model deployments in medical imaging. FUNDING: National Institute of Biomedical Imaging and Bioengineering, MIDRC grant of National Institutes of Health, US National Science Foundation, National Library of Medicine of the National Institutes of Health, and Taiwan Ministry of Science and Technology.

Validation and algorithmic audit of a deep learning system for the detection of proximal femoral fractures in patients in the emergency department: a diagnostic accuracy study.

BACKGROUND: Proximal femoral fractures are an important clinical and public health issue associated with substantial morbidity and early mortality. Artificial intelligence might offer improved diagnostic accuracy for these fractures, but typical approaches to testing of artificial intelligence models can underestimate the risks of artificial intelligence-based diagnostic systems. METHODS: We present a preclinical evaluation of a deep learning model intended to detect proximal femoral fractures in frontal x-ray films in emergency department patients, trained on films from the Royal Adelaide Hospital (Adelaide, SA, Australia). This evaluation included a reader study comparing the performance of the model against five radiologists (three musculoskeletal specialists and two general radiologists) on a dataset of 200 fracture cases and 200 non-fractures (also from the Royal Adelaide Hospital), an external validation study using a dataset obtained from Stanford University Medical Center, CA, USA, and an algorithmic audit to detect any unusual or unexpected model behaviour. FINDINGS: In the reader study, the area under the receiver operating characteristic curve (AUC) for the performance of the deep learning model was 0·994 (95% CI 0·988-0·999) compared with an AUC of 0·969 (0·960-0·978) for the five radiologists. This strong model performance was maintained on external validation, with an AUC of 0·980 (0·931-1·000). However, the preclinical evaluation identified barriers to safe deployment, including a substantial shift in the model operating point on external validation and an increased error rate on cases with abnormal bones (eg, Paget's disease). INTERPRETATION: The model outperformed the radiologists tested and maintained performance on external validation, but showed several unexpected limitations during further testing. Thorough preclinical evaluation of artificial intelligence models, including algorithmic auditing, can reveal unexpected and potentially harmful behaviour even in high-performance artificial intelligence systems, which can inform future clinical testing and deployment decisions. FUNDING: None.

The changing profile of obstructive sleep apnea: long term trends in characteristics of patients presenting for diagnostic polysomnography.

Introduction: We aimed to analyze long-term trends in characteristics of patients undergoing diagnostic polysomnography (PSG) and subsequently diagnosed with obstructive sleep apnea (OSA) to inform delivery of sleep services. Material and Methods: We studied 24,510 consecutive patients undergoing PSG at a tertiary-care sleep service between 1989 and 2013. OSA was defined by an apnea hypopnea index (AHI)≥ 5 events/hour. Changes to hypopnea definition and flow sensing techniques in 2002 created two distinct AHI scoring periods: American Sleep Disorders Association (ASDA) 1989 - July 2002 and American Academy of Sleep Medicine (Chicago) from August 2002. Results: Over 23.5 years there was a steady increase in proportion of females (15% to 45%), small increases in average age and BMI, and a small decline in socioeconomic status in the overall group. AHI varied between scoring periods both overall [ASDA 10.8/h (3.2-29.6), Chicago 24.3/h (11.8-48.1)] and in the large subgroup (80.7%) diagnosed with OSA [ASDA 20.7/h (10.6-44.1), Chicago 27.4/h (14.8-51.5)]. OSA diagnosis rates increased in the Chicago period (ASDA 66%, Chicago 91%). Increases in AHI and proportion diagnosed appeared better explained by changes in scoring methods than key OSA risk factors. Conclusion: Temporal increases in proportion of females and decreases in socioeconomic status of people undergoing PSG may reflect greater community awareness of sleep disorders. Temporal increases in age and obesity are consistent with secular trends. Changes in scoring methods have major impacts on OSA diagnosis and judgement of disease severity, with important implications for contemporary resourcing of sleep services and interpretation of historical OSA data.

The Epidemiology and Burden of Childhood Chronic Pancreatitis in South Australia.

OBJECTIVE: To assess longitudinal, population-based data on the prevalence and impact of chronic pancreatitis in children. STUDY DESIGN: Administrative data linkage was used to ascertain an index cohort consisting of all individuals who had an initial diagnosis of chronic pancreatitis before age 19 years in the South Australian public hospital system between June 2000 and June 2019. Age- and sex-matched controls were drawn from the general population of South Australia, children with type 1 diabetes, and children with type 2 diabetes. Main outcomes and measures included hospital visits, days in hospital, emergency department (ED) visits, intensive care unit (ICU) admissions, education comparators, and incidence and prevalence estimates. RESULTS: A total of 73 incident cases were identified. The crude prevalence and incidence of pediatric chronic pancreatitis were estimated at 6.8/100 000 and 0.98/100 000 per year, respectively. Of the index cohort, 24 cases (32.8%) of pediatric chronic pancreatitis were identified as occurring in children of Aboriginal and/or Torres Strait Islander descent. Compared with matched general population controls, children with chronic pancreatitis averaged 11-fold more hospital visits, 5-fold more ED visits, and 9-fold more ICU admissions; spent 10-fold more days in the hospital; and had a 2-fold higher rate of absence from school (P < .001 for all). Similarly, children with chronic pancreatitis used substantially more health resources than children with type 1 or 2 diabetes. CONCLUSIONS: Pediatric patients with chronic pancreatitis consume a high volume of public health services and are significantly impacted in their ability to engage in education.

Whole-genome association analyses of sleep-disordered breathing phenotypes in the NHLBI TOPMed program.

BACKGROUND: Sleep-disordered breathing is a common disorder associated with significant morbidity. The genetic architecture of sleep-disordered breathing remains poorly understood. Through the NHLBI Trans-Omics for Precision Medicine (TOPMed) program, we performed the first whole-genome sequence analysis of sleep-disordered breathing. METHODS: The study sample was comprised of 7988 individuals of diverse ancestry. Common-variant and pathway analyses included an additional 13,257 individuals. We examined five complementary traits describing different aspects of sleep-disordered breathing: the apnea-hypopnea index, average oxyhemoglobin desaturation per event, average and minimum oxyhemoglobin saturation across the sleep episode, and the percentage of sleep with oxyhemoglobin saturation < 90%. We adjusted for age, sex, BMI, study, and family structure using MMSKAT and EMMAX mixed linear model approaches. Additional bioinformatics analyses were performed with MetaXcan, GIGSEA, and ReMap. RESULTS: We identified a multi-ethnic set-based rare-variant association (p = 3.48 × 10-8) on chromosome X with ARMCX3. Additional rare-variant associations include ARMCX3-AS1, MRPS33, and C16orf90. Novel common-variant loci were identified in the NRG1 and SLC45A2 regions, and previously associated loci in the IL18RAP and ATP2B4 regions were associated with novel phenotypes. Transcription factor binding site enrichment identified associations with genes implicated with respiratory and craniofacial traits. Additional analyses identified significantly associated pathways. CONCLUSIONS: We have identified the first gene-based rare-variant associations with objectively measured sleep-disordered breathing traits. Our results increase the understanding of the genetic architecture of sleep-disordered breathing and highlight associations in genes that modulate lung development, inflammation, respiratory rhythmogenesis, and HIF1A-mediated hypoxic response.

Australian experience with total pancreatectomy with islet autotransplantation to treat chronic pancreatitis.

BACKGROUND: This study aimed to describe the clinical outcomes of total pancreatectomy with islet autotransplantation (TP-IAT) in Australia. METHODS: Individuals selected for TP-IAT surgery according to the Minnesota Criteria (Appendix) without evidence of diabetes were evaluated including time to transplantation from pancreatectomy, islet numbers infused and post-transplantation HbA1c, C-peptide, total daily insulin and analgesic requirement. RESULTS: Sixteen individuals underwent TP-IAT from Australia and New Zealand between 2010 and 2020. Two recipients are deceased. The median islet equivalents/kg infused was 4244 (interquartile range (IQR) 2290-7300). The median C-peptide 1 month post-TP-IAT was 384 (IQR 210-579) pmol/L and at median 29.5 (IQR 14.5-46.5) months from transplant was 395 (IQR 139-862) pmol/L. Insulin independence was achieved in eight of 15 (53.3%) surviving recipients. A higher islet equivalents transplanted was most strongly associated with the likelihood of insulin independence (P < 0.05). Of the 15 surviving recipients, 14 demonstrated substantial reduction in analgesic requirement. CONCLUSION: The TP-IAT programme in Australia has been a successful new therapy for the management of individuals with chronic pancreatitis including hereditary forms refractory to medical treatment to improve pain management with 50% insulin independence rates.

A survey of clinicians on the use of artificial intelligence in ophthalmology, dermatology, radiology and radiation oncology.

Artificial intelligence technology has advanced rapidly in recent years and has the potential to improve healthcare outcomes. However, technology uptake will be largely driven by clinicians, and there is a paucity of data regarding the attitude that clinicians have to this new technology. In June-August 2019 we conducted an online survey of fellows and trainees of three specialty colleges (ophthalmology, radiology/radiation oncology, dermatology) in Australia and New Zealand on artificial intelligence. There were 632 complete responses (n = 305, 230, and 97, respectively), equating to a response rate of 20.4%, 5.1%, and 13.2% for the above colleges, respectively. The majority (n = 449, 71.0%) believed artificial intelligence would improve their field of medicine, and that medical workforce needs would be impacted by the technology within the next decade (n = 542, 85.8%). Improved disease screening and streamlining of monotonous tasks were identified as key benefits of artificial intelligence. The divestment of healthcare to technology companies and medical liability implications were the greatest concerns. Education was identified as a priority to prepare clinicians for the implementation of artificial intelligence in healthcare. This survey highlights parallels between the perceptions of different clinician groups in Australia and New Zealand about artificial intelligence in medicine. Artificial intelligence was recognized as valuable technology that will have wide-ranging impacts on healthcare.

Continuous positive airway pressure treatment, glycemia, and diabetes risk in obstructive sleep apnea and comorbid cardiovascular disease

OBJECTIVE: Despite evidence of a relationship among obstructive sleep apnea (OSA), metabolic dysregulation, and diabetes, it is uncertain whether OSA treatment can improve metabolic parameters. We sought to determine effects of long-term continuous positive airway pressure (CPAP) treatment on glycemic control and diabetes risk in patients with cardiovascular disease (CVD) and OSA. RESEARCH DESIGN AND METHODS: Blood, medical history, and personal data were collected in a substudy of 888 participants in the Sleep Apnea cardioVascular Endpoints (SAVE) trial in which patients with OSA and stable CVD were randomized to receive CPAP plus usual care, or usual care alone. Serum glucose and glycated hemoglobin A1c (HbA1c) were measured at baseline, 6 months, and 2 and 4 years and incident diabetes diagnoses recorded. RESULTS: Median follow-up was 4.3 years. In those with preexisting diabetes (n = 274), there was no significant difference between the CPAP and usual care groups in serum glucose, HbA1c, or antidiabetic medications during follow-up. There were also no significant between-group differences in participants with prediabetes (n = 452) or new diagnoses of diabetes. Interaction testing suggested that women with diabetes did poorly in the usual care group, while their counterparts on CPAP therapy remained stable. CONCLUSIONS: Among patients with established CVD and OSA, we found no evidence that CPAP therapy over several years affects glycemic control in those with diabetes or prediabetes or diabetes risk over standard-of-care treatment. The potential differential effect according to sex deserves further investigation.

Definition and diagnosis of cerebral palsy in genetic studies: a systematic review.

AIM: To conduct a systematic review of phenotypic definition and case ascertainment in published genetic studies of cerebral palsy (CP) to inform guidelines for the reporting of such studies. METHOD: Inclusion criteria comprised genetic studies of candidate genes, with CP as the outcome, published between 1990 and 2019 in the PubMed, Embase, and BIOSIS Citation Index databases. RESULTS: Fifty-seven studies met the inclusion criteria. We appraised how CP was defined, the quality of information on case ascertainment, and compliance with international consensus guidelines. Seven studies (12%) were poorly described, 33 studies (58%) gave incomplete information, and 17 studies (30%) were well described. Missing key information precluded determining how many studies complied with the definition by Rosenbaum et al. Only 18 out of 57 studies (32%) were compliant with the Surveillance of Cerebral Palsy in Europe (SCPE) international guidelines on defining CP. INTERPRETATION: Limited compliance with international consensus guidelines on phenotypic definition and mediocre reporting of CP case ascertainment hinders the comparison of results among genetic studies of CP (including meta-analyses), thereby limiting the quality, interpretability, and generalizability of study findings. Compliance with the SCPE guidelines is important for ongoing gene discovery efforts in CP, given the potential for misclassification of unrelated neurological conditions as CP.

The Relationship of Sleep Duration with Ethnicity and Chronic Disease in a Canadian General Population Cohort.

STUDY OBJECTIVES: Sleep duration is an important marker of sleep quality and overall sleep health. Both too little and too much sleep are associated with poorer health outcomes. We hypothesized that ethnicity-specific differences in sleep duration exist. METHODS: This cross-sectional study utilized questionnaire data from the Ontario Health Study (OHS), a multi-ethnic population-based cohort of Canadian adult residents aged 18 to 99 years, who provided medical, socio-demographic, and sleep information. Generalised linear models were used to investigate the association of sleep duration with ethnicity. RESULTS: The study sample consisted of 143,307 adults (60.4% women). The sample was multi-ethnic, including self-identified Aboriginal, Arab, Black, Chinese, Filipino, Hispanic, Japanese, Korean, Mixed (>1 ethnicity), South Asian, South-East Asian, West Asian, and White ethnicities. Univariate analyses found that mean sleep duration compared to the White reference group (7.34 hours) was shorter in the Filipino (6.93 hours, 25 min less), Black (6.96 hours, 23 min less), Japanese (7.02 hours, 19 min less), Chinese (7.23 hours, 7 min less), and Mixed (7.27 hours, 4 min less) groups (all P<0.001). Mean sleep duration was shorter in men (7.25 hours) compared to women (7.37 hours) in the cohort as a whole (P<0.001), and in all ethnic groups (P<0.001). Multivariate analyses, adjusted for a wide range of potential risk factors, and analysis of sleep duration as a categorical variable ("short", "average", and "long" sleepers) confirmed these relationships. Both sleep duration and ethnicity were independent significant predictors of a range of physician-diagnosed morbidities including diabetes, stroke, and depression. CONCLUSION: Important differences exist in sleep duration between ethnic groups and may contribute to observed health disparities. Our results highlight the need for ethnicity-specific targeted education on the importance of prioritizing sleep for good health, and the need to account appropriately for ethnicity in future epidemiological, clinical, and translational research into sleep and related conditions.