Search

Background: Delays in patient flow and a shortage of hospital beds are commonplace in hospitals during periods of increased infection incidence, such as seasonal influenza and the COVID-19 pandemic. The objective of this study was to develop and evaluate the efficacy of machine learning methods at identifying and ranking the real-time readiness of individual patients for discharge, with the goal of improving patient flow within hospitals during periods of crisis., Methods and Performance: Electronic Health Record data from Oxford University Hospitals was used to train independent models to classify and rank patients' real-time readiness for discharge within 24 hours, for patient subsets according to the nature of their admission (planned or emergency) and the number of days elapsed since their admission. A strategy for the use of the models' inference is proposed, by which the model makes predictions for all patients in hospital and ranks them in order of likelihood of discharge within the following 24 hours. The 20% of patients with the highest ranking are considered as candidates for discharge and would therefore expect to have a further screening by a clinician to confirm whether they are ready for discharge or not. Performance was evaluated in terms of positive predictive value (PPV), i.e., the proportion of these patients who would have been correctly deemed as 'ready for discharge' after having the second screening by a clinician. Performance was high for patients on their first day of admission (PPV = 0.96/0.94 for planned/emergency patients respectively) but dropped for patients further into a longer admission (PPV = 0.66/0.71 for planned/emergency patients still in hospital after 7 days)., Conclusion: We demonstrate the efficacy of machine learning methods at making operationally focused, next-day discharge readiness predictions for all individual patients in hospital at any given moment and propose a strategy for their use within a decision-support tool during crisis periods., Competing Interests: DWE has received lecture fees and conference expenses from Gilead. No other author has a conflict of interest to declare. This does not alter our adherence to PLOS ONE policies on sharing data and materials.

Background: Diabetic retinopathy (DR) can lead to visual impairment and blindness if not detected and treated in time. Knowing the barriers/enablers in advance in contrasting different country income settings may accelerate development of a successful DR screening (DRS) program. This would be especially applicable in the low-income settings with the rising prevalence of DR., Objectives: The aim of this systematic review is to identify and contrast the barriers/enablers to DRS for different contexts using both consumers i.e., people with diabetes (PwDM) and provider perspectives and system level factors in different country income settings., Methods: We searched MEDLINE, Embase, CENTRAL in the Cochrane Library from the databases start date to December 2018. We included the studies reported on barriers and enablers to access DRS services based at health care facilities. We categorised and synthesized themes related to the consumers (individuals), providers and the health systems (environment) as main dimensions according to the constructs of social cognitive theory, supported by the quantitative measures i.e., odds ratios as reported by each of the study authors., Main Results: We included 77 studies primarily describing the barriers and enablers. Most of the studies were from high income settings (72.7%, 56/77) and cross sectional in design (76.6%, 59/77). From the perspectives of consumers, lack of knowledge, attitude, awareness and motivation were identified as major barriers. The enablers were fear of blindness, proximity of screening facility, experiences of vision loss and being concerned of eye complications. In providers' perspectives, lack of skilled human resources, training programs, infrastructure of retinal imaging and cost of services were the main barriers. Higher odds of uptake of DRS services was observed when PwDM were provided health education (odds ratio (OR) 4.3) and having knowledge on DR (OR range 1.3-19.7)., Conclusion: Knowing the barriers to access DRS is a pre-requisite in development of a successful screening program. The awareness, knowledge and attitude of the consumers, availability of skilled human resources and infrastructure emerged as the major barriers to access to DRS in any income setting., Competing Interests: The authors have declared that no competing interests exist.

Purpose: To describe an approach to the use of optical coherence tomography (OCT) imaging in large, population-based studies, including methods for OCT image acquisition, storage, and the remote, rapid, automated analysis of retinal thickness., Methods: In UK Biobank, OCT images were acquired between 2009 and 2010 using a commercially available "spectral domain" OCT device (3D OCT-1000, Topcon). Images were obtained using a raster scan protocol, 6 mm x 6 mm in area, and consisting of 128 B-scans. OCT image sets were stored on UK Biobank servers in a central repository, adjacent to high performance computers. Rapid, automated analysis of retinal thickness was performed using custom image segmentation software developed by the Topcon Advanced Biomedical Imaging Laboratory (TABIL). This software employs dual-scale gradient information to allow for automated segmentation of nine intraretinal boundaries in a rapid fashion., Results: 67,321 participants (134,642 eyes) in UK Biobank underwent OCT imaging of both eyes as part of the ocular module. 134,611 images were successfully processed with 31 images failing segmentation analysis due to corrupted OCT files or withdrawal of subject consent for UKBB study participation. Average time taken to call up an image from the database and complete segmentation analysis was approximately 120 seconds per data set per login, and analysis of the entire dataset was completed in approximately 28 days., Conclusions: We report an approach to the rapid, automated measurement of retinal thickness from nearly 140,000 OCT image sets from the UK Biobank. In the near future, these measurements will be publically available for utilization by researchers around the world, and thus for correlation with the wealth of other data collected in UK Biobank. The automated analysis approaches we describe may be of utility for future large population-based epidemiological studies, clinical trials, and screening programs that employ OCT imaging., Competing Interests: Charles A. Reisman and Qi Yang are both employed by Topcon Medical Systems, Inc. Other than this employment relationship, Topcon has not funded or sponsored this research in any way and furthermore, this relationship does not affect or alter the authors' adherence to PLOS journals’ data sharing or materials policies.

Objective: Digital retinal imaging is an established method of screening for diabetic retinopathy (DR). It has been established that currently about 1% of the world's blind or visually impaired is due to DR. However, the increasing prevalence of diabetes mellitus and DR is creating an increased workload on those with expertise in grading retinal images. Safe and reliable automated analysis of retinal images may support screening services worldwide. This study aimed to compare the Iowa Detection Program (IDP) ability to detect diabetic eye diseases (DED) to human grading carried out at Moorfields Reading Centre on the population of Nakuru Study from Kenya., Participants: Retinal images were taken from participants of the Nakuru Eye Disease Study in Kenya in 2007/08 (n = 4,381 participants [NW6 Topcon Digital Retinal Camera])., Methods: First, human grading was performed for the presence or absence of DR, and for those with DR this was sub-divided in to referable or non-referable DR. The automated IDP software was deployed to identify those with DR and also to categorize the severity of DR., Main Outcome Measures: The primary outcomes were sensitivity, specificity, and positive and negative predictive value of IDP versus the human grader as reference standard., Results: Altogether 3,460 participants were included. 113 had DED, giving a prevalence of 3.3% (95% CI, 2.7-3.9%). Sensitivity of the IDP to detect DED as by the human grading was 91.0% (95% CI, 88.0-93.4%). The IDP ability to detect DED gave an AUC of 0.878 (95% CI 0.850-0.905). It showed a negative predictive value of 98%. The IDP missed no vision threatening retinopathy in any patients and none of the false negative cases met criteria for treatment., Conclusions: In this epidemiological sample, the IDP's grading was comparable to that of human graders'. It therefore might be feasible to consider inclusion into usual epidemiological grading.

Purpose: To examine the cross sectional and longitudinal relationship between cardiovascular risk factors and age-related macular degeneration (AMD) in a large British cohort study., Methods: The EPIC Norfolk Eye study is nested in a larger prospective cohort study. Data on cardiovascular risk factors were collected at baseline (1993-1997) and follow up (2006-2011) via clinical examination, validated lifestyle questionnaires and serum blood samples. AMD was ascertained using standardised grading of fundus photographs at the follow up. Logistic regression was used to examine associations between baseline and follow up risk factors with AMD., Results: 5,344 pairs (62.0% of total 8623) of fundus photographs were of sufficient quality for grading of AMD in participants with mean age of 67.4 years old (range 44-91) at diagnosis. There were 28 cases of late AMD (0.5%, 95% confidence interval (CI)=0.3-0.8%) and 645 cases of early AMD (12.1%, 95%CI=11.2-13.0.%). In multivariable analysis, older people with higher levels of baseline high density lipoprotein- cholesterol (HDL-C ) and C-reactive protein (CRP) were more likely to have any signs of AMD, after adjusting for sex, education, smoking, and systolic blood pressure. In cross sectional analysis, only older age and higher HDL were significantly associated with AMD., Conclusions: We have found that older age and higher levels of CRP and HDL-C were associated with increased odds of AMD in this population in the longitudinal analysis, but older age and HDL-C, not CRP was significantly associated with AMD in the cross sectional analysis. The prevalence of AMD in this cohort was low compared to other cohorts in Europe, the US and Australia, and probably reflects the some selection biases in follow up participation as well as the low rate of smoking among our healthy participants.

Objective: Diabetic retinopathy (DR) is the leading cause of blindness in people of working age. The purpose of this paper is to report the prevalence and cardiovascular associations of diabetic retinopathy and maculopathy (DMac) in Germany., Research Design and Methods: The Gutenberg Health Study (GHS) is a population-based study with 15,010 participants aged between 35 at 74 years from the city of Mainz and the district of Mainz-Bingen. We determined the weighted prevalence of DR and DMac by assessing fundus photographs of persons with diabetes from the GHS data base. Diabetes was defined as HbA1c ≥ 6.5%, known diagnosis diabetes mellitus or known diabetes medication. Furthermore, we analysed the association between DR and cardiovascular risk factors and diseases., Results: Overall, 7.5% (1,124/15,010) of the GHS cohort had diabetes. Of these, 27.7% were unaware of their disease and thus were newly diagnosed by their participation in the GHS. The prevalence of DR and DMac was 21.7% and 2.3%, respectively among patients with diabetes. Vision-threatening disease was present in 5% of the diabetic cohort. In the multivariable analysis DR (all types) was associated with age (Odds Ratio [95% confidence interval]: 0.97 [0.955-0.992]; p = 0.006) arterial hypertension (1.90 [1.190-3.044]; p = 0.0072) and vision-threatening DR with obesity (3.29 [1.504-7.206]; p = 0.0029). DR (all stages) and vision-threatening DR were associated with duration of diabetes (1.09 [1.068-1.114]; p<0.0001 and 1.18 [1.137-1.222]; p<0.0001, respectively)., Conclusions: Our calculations suggest that approximately 142 000 persons aged between 35 and 74 years have vision threatening diabetic retinal disease in Germany [corrected].Prevalence of DR was lower in the GHS compared to East-Asian studies. Associations were found with age, arterial hypertension, obesity, and duration of diabetes mellitus.

Purpose: This paper describes the study design, methodology, cohort profile and self-reported diseases in the ophthalmological branch of the Gutenberg Health Study (GHS)., Methods: The GHS is an ongoing, prospective, interdisciplinary, single-center, population-based cohort study in Germany. The main goals of the ophthalmological section are to assess the prevalence and incidence of ocular diseases and to explore risk factors, genetic determinants and associations with systemic diseases and conditions. The eye examination at baseline included a medical history, self-reported eye diseases, visual acuity, refractive errors, intraocular pressure, visual field, pachymetry, keratometry, fundus photography and tear sampling. The 5-year follow-up visit additionally encompassed optical coherence tomography, anterior segment imaging and optical biometry. The general examination included anthropometry; blood pressure measurement; carotid artery ultrasound; electrocardiogram; echocardiography; spirometry; cognitive tests; questionnaires; assessment of mental conditions; and DNA, RNA, blood and urine sampling., Results: Of 15,010 participants (aged 35-74 years at the time of inclusion), ocular data are available for 14,700 subjects (97.9%). The mean visual acuity (standard deviation), mean spherical equivalent, median decimal visual acuity, and mean intraocular pressure were 0.08 (0.17) logMar, -0.42 (2.43) diopters, 0.9 and 14.24 (2.79) mm Hg, respectively. The frequencies of self-reported strabismus, glaucoma, surgery for retinal detachment and retinal vascular occlusions were 2.7%, 2.3%, 0.2% and 0.4%, respectively., Conclusions: The GHS is the most extensive dataset of ophthalmic diseases and conditions and their risk factors in Germany and one of the largest cohorts worldwide. This dataset will provide new insight in the epidemiology of ophthalmic diseases and related medical specialties.

Aim: Crowdsourcing is the process of simplifying and outsourcing numerous tasks to many untrained individuals. Our aim was to assess the performance and repeatability of crowdsourcing in the classification of normal and glaucomatous discs from optic disc images., Methods: Optic disc images (N = 127) with pre-determined disease status were selected by consensus agreement from grading experts from a large cohort study. After reading brief illustrative instructions, we requested that knowledge workers (KWs) from a crowdsourcing platform (Amazon MTurk) classified each image as normal or abnormal. Each image was classified 20 times by different KWs. Two study designs were examined to assess the effect of varying KW experience and both study designs were conducted twice for consistency. Performance was assessed by comparing the sensitivity, specificity and area under the receiver operating characteristic curve (AUC)., Results: Overall, 2,540 classifications were received in under 24 hours at minimal cost. The sensitivity ranged between 83-88% across both trials and study designs, however the specificity was poor, ranging between 35-43%. In trial 1, the highest AUC (95%CI) was 0.64(0.62-0.66) and in trial 2 it was 0.63(0.61-0.65). There were no significant differences between study design or trials conducted., Conclusions: Crowdsourcing represents a cost-effective method of image analysis which demonstrates good repeatability and a high sensitivity. Optimisation of variables such as reward schemes, mode of image presentation, expanded response options and incorporation of training modules should be examined to determine their effect on the accuracy and reliability of this technique in retinal image analysis.

Background: To describe structural and functional changes associated with diabetic macular oedema (DMO) treated with intravitreal bevacizumab over 24 months., Methods: A post-hoc analysis of the data of 34 patients that completed 24 months follow-up in the intravitreal bevacizumab arm of a prospective, randomized controlled trial (BOLT study) was performed. The outcome measures previously used in clinical trials of intravitreal ranibizumab in DMO were employed to describe the visual acuity and macular thickness changes at 12 and 24 months., Results: The standard outcomes of mean change in best corrected visual acuity (BCVA) and central macular thickness (CMT) in participants treated with bevacizumab were comparable to those reported in association with ranibizumab. However, exploratory analyses showed that thick maculae at baseline defined as CMT of ≥ 400 µm, remained significantly thicker than those <400 µm with intensive bevacizumab therapy, despite a comparable gain in visual acuity at both 12 and 24 months. The proportion of subjects that attained a dry macula doubled in both CMT groups between the 12 and 24-month time-points., Conclusions: These findings provide valuable information both for clinical practice and trials. Further studies are required to investigate the impact of intravitreal bevacizumab on retinal thickness profiles in DMO.

Aim: Crowdsourcing is the process of outsourcing numerous tasks to many untrained individuals. Our aim was to assess the performance and repeatability of crowdsourcing for the classification of retinal fundus photography., Methods: One hundred retinal fundus photograph images with pre-determined disease criteria were selected by experts from a large cohort study. After reading brief instructions and an example classification, we requested that knowledge workers (KWs) from a crowdsourcing platform classified each image as normal or abnormal with grades of severity. Each image was classified 20 times by different KWs. Four study designs were examined to assess the effect of varying incentive and KW experience in classification accuracy. All study designs were conducted twice to examine repeatability. Performance was assessed by comparing the sensitivity, specificity and area under the receiver operating characteristic curve (AUC)., Results: Without restriction on eligible participants, two thousand classifications of 100 images were received in under 24 hours at minimal cost. In trial 1 all study designs had an AUC (95%CI) of 0.701(0.680-0.721) or greater for classification of normal/abnormal. In trial 1, the highest AUC (95%CI) for normal/abnormal classification was 0.757 (0.738-0.776) for KWs with moderate experience. Comparable results were observed in trial 2. In trial 1, between 64-86% of any abnormal image was correctly classified by over half of all KWs. In trial 2, this ranged between 74-97%. Sensitivity was ≥ 96% for normal versus severely abnormal detections across all trials. Sensitivity for normal versus mildly abnormal varied between 61-79% across trials., Conclusions: With minimal training, crowdsourcing represents an accurate, rapid and cost-effective method of retinal image analysis which demonstrates good repeatability. Larger studies with more comprehensive participant training are needed to explore the utility of this compelling technique in large scale medical image analysis.

In any diabetic retinopathy screening program, about two-thirds of patients have no retinopathy. However, on average, it takes a human expert about one and a half times longer to decide an image is normal than to recognize an abnormal case with obvious features. In this work, we present an automated system for filtering out normal cases to facilitate a more effective use of grading time. The key aim with any such tool is to achieve high sensitivity and specificity to ensure patients' safety and service efficiency. There are many challenges to overcome, given the variation of images and characteristics to identify. The system combines computed evidence obtained from various processing stages, including segmentation of candidate regions, classification and contextual analysis through Hidden Markov Models. Furthermore, evolutionary algorithms are employed to optimize the Hidden Markov Models, feature selection and heterogeneous ensemble classifiers. In order to evaluate its capability of identifying normal images across diverse populations, a population-oriented study was undertaken comparing the software's output to grading by humans. In addition, population based studies collect large numbers of images on subjects expected to have no abnormality. These studies expect timely and cost-effective grading. Altogether 9954 previously unseen images taken from various populations were tested. All test images were masked so the automated system had not been exposed to them before. This system was trained using image subregions taken from about 400 sample images. Sensitivities of 92.2% and specificities of 90.4% were achieved varying between populations and population clusters. Of all images the automated system decided to be normal, 98.2% were true normal when compared to the manual grading results. These results demonstrate scalability and strong potential of such an integrated computational intelligence system as an effective tool to assist a grading service.

Objectives: To determine the duration of protection from hepatitis B vaccine given in infancy and early childhood and asses risk factors for HBV infection and chronic infection., Methods: In 1984 infant HBV vaccination was started in two Gambian villages. Cross sectional serological surveys have been undertaken every 4 years to determine vaccine efficacy. In the current survey 84.6% of 1508 eligible participants aged 1-28 years were tested. A spouse study was conducted in females (aged 14 years and above) and their male partners., Results: Vaccine efficacy against chronic infection with hepatitis B virus was 95.1% (95% confidence interval 91.5% to 97.1%), which did not vary significantly between age groups or village. Efficacy against infection was 85.4% (82.7% to 87.7%), falling significantly with age. Concentrations of hepatitis B antibody fell exponentially with age varying according to peak response: 20 years after vaccination only 17.8% (95% CI 10.1-25.6) of persons with a low peak response (10-99 mIU/ml) had detectable HBs antibody compared to 27% (21.9% to 32.2%) of those with a high peak response (>999 mIU/ml). Time since vaccination and a low peak response were the strongest risk factors for HBV infections; males were more susceptible, marriage was not a significant risk for females. Hepatitis B DNA was not detected after infection, which tested soley core antibody positive. An undetectable peak antibody response of <10 mIU/ml and a mother who was hepatitis B e antigen positive were powerful risk factors for chronic infection., Conclusions: Adolescents and young adults vaccinated in infancy are at increased risk of hepatitis B infection, but not chronic infection. Married women were not at increased risk. There is no compelling evidence for the use of a booster dose of HBV vaccine in The Gambia.

Objectives: HIV infection is associated with an increased risk of age-related morbidity mediated by immune dysfunction, atherosclerosis and inflammation. Changes in retinal vessel calibre may reflect cumulative structural damage arising from these mechanisms. The relationship of retinal vessel calibre with clinical and demographic characteristics was investigated in a population of HIV-infected individuals in South Africa., Methods: Case-control study of 491 adults ≥30 years, composed of 242 HIV-infected adults and 249 age- and gender-matched HIV-negative controls. Retinal vessel calibre was measured using computer-assisted techniques to determine mean arteriolar and venular diameters of each eye., Results: The median age was 40 years (IQR: 35-48 years). Among HIV-infected adults, 87.1% were receiving highly active antiretroviral therapy (HAART) (median duration, 58 months), their median CD4 count was 468 cells/µL, and 84.3% had undetectable plasma viral load. Unadjusted mean retinal arteriolar diameters were 163.67±17.69 µm in cases and 161.34±17.38 µm in controls (p = 0.15). Unadjusted mean venular diameters were 267.77±18.21 µm in cases and 270.81±18.98 µm in controls (p = 0.07). Age modified the effect of retinal arteriolar and venular diameters in relation to HIV status, with a tendency towards narrower retinal diameters in HIV cases but not in controls. Among cases, retinal arteriolar diameters narrowed with increasing duration of HAART, independently of age (167.83 µm <3 years of HAART vs. 158.89 µm >6 years, p-trend = 0.02), and with a HIV viral load >10,000 copies/mL while on HAART (p = 0.05). HIV-related venular changes were not detected., Conclusions: Narrowing of retinal arteriolar diameters is associated with HAART duration and viral load, and may reflect heightened inflammatory and pro-atherogenic states of the systemic vasculature. Measurement of retinal vascular calibre could be an innovative non-invasive method of estimating vascular risk in HIV-infected individuals.

Macular Telangiectasia type 2 (MacTel) is a relatively rare macular disease of adult onset presenting with distortions in the visual field and leading to progressive loss of visual acuity. For the purpose of a gene mapping study, several pedigrees were ascertained with multiple affected family members. Seventeen families with a total of 71 individuals (including 45 affected or possibly affected) were recruited at clinical centers in 7 countries under the auspices of the MacTel Project. The disease inheritance was consistent with autosomal dominant segregation with reduced penetrance. Genome-wide linkage analysis was performed, followed by analysis of recombination breakpoints. Linkage analysis identified a single peak with multi-point LOD score of 3.45 on chromosome 1 at 1q41-42 under a dominant model. Recombination mapping defined a minimal candidate region of 15.6 Mb, from 214.32 (rs1579634; 219.96 cM) to 229.92 Mb (rs7542797; 235.07 cM), encompassing the 1q41-42 linkage peak. Sanger sequencing of the top 14 positional candidates genes under the linkage peak revealed no causal variants in these pedigrees.

Objectives: Preparatory to a community trial investigating how best to deliver rectal artesunate as pre-referral treatment for severe malaria; local understanding, perceptions of signs/symptoms of severe malaria and treatment-seeking patterns for and barriers to seeking biomedical treatment were investigated., Methodology/principal Findings: 19 key informant interviews, 12 in-depth interviews and 14 focus group discussions targeting care-givers, opinion leaders, and formal and informal health care providers were conducted. Monthly fever episodes and danger signs or symptoms associated with severe malaria among under-fives were recorded. Respondents recognized convulsions, altered consciousness and coma, and were aware of their risks if not treated. But, these symptoms were perceived to be caused by supernatural forces, and traditional healers were identified as primary care providers. With some delay, mothers eventually visited a health facility when convulsions were part of the illness, despite pressures against this. Although vomiting and failure to eat/suck/drink were associated with malaria, they were not considered as indicators of danger signs unless combined with another more severe symptom. Study communities were familiar with rectal application of medicines., Conclusions/significance: Communities' recognition and awareness of major symptoms of severe malaria could encourage action, but perceptions of their causes and poor discrimination of other danger signs - vomiting and failure to feed - might impede early treatment. An effective health education targeting parents/guardians, decision-makers/advisors, and formal and informal care providers might be a prerequisite for successful introduction of rectal artemisinins as an emergency treatment. Role of traditional healers in delivering such medication to the community should be explored.