Detecting diabetic retinopathy through machine learning on electronic health record data from an urban, safety net healthcare system.
Ontology highlight
ABSTRACT: Objective
Clinical guidelines recommend annual eye examinations to detect diabetic retinopathy (DR) in patients with diabetes. However, timely DR detection remains a problem in medically underserved and under-resourced settings in the United States. Machine learning that identifies patients with latent/undiagnosed DR could help to address this problem.Materials and methods
Using electronic health record data from 40 631 unique diabetic patients seen at Los Angeles County Department of Health Services healthcare facilities between January 1, 2015 and December 31, 2017, we compared ten machine learning environments, including five classifier models, for assessing the presence or absence of DR. We also used data from a distinct set of 9300 diabetic patients seen between January 1, 2018 and December 31, 2018 as an external validation set.Results
Following feature subset selection, the classifier with the best AUC on the external validation set was a deep neural network using majority class undersampling, with an AUC of 0.8, the sensitivity of 72.17%, and specificity of 74.2%.Discussion
A deep neural network produced the best AUCs and sensitivity results on the test set and external validation set. Models are intended to be used to screen guideline noncompliant diabetic patients in an urban safety-net setting.Conclusion
Machine learning on diabetic patients' routinely collected clinical data could help clinicians in safety-net settings to identify and target unscreened diabetic patients who potentially have undiagnosed DR.
SUBMITTER: Ogunyemi OI
PROVIDER: S-EPMC8374369 | biostudies-literature |
REPOSITORIES: biostudies-literature
Similar Datasets
Project description:This study sought to assess the association between axial length (AL) and diabetic retinopathy (DR) in a diverse cohort of patients and to investigate the impact of sex on this relationship. An urban safety net hospital database was used for this cross-sectional observational study. Diabetic patients who underwent fundus photography and AL measurement between March 2017 and June 2020 were included. The fundus photographs were graded following the Early Treatment of Diabetic Retinopathy Study criteria. The study enrolled 1843 patients with diabetes (mean age: 56.9 ± 12.1 years; AL: 23.56 ± 1.12 mm), including 931 men and 912 women. Male sex was a risk factor for diabetic retinopathy (P = 0.001; odds ratio [OR] 1.5, 95% confidence interval [CI] 1.18-1.98). A higher DR prevalence was associated with shorter AL both in men (P = 0.003; OR 0.77; 95% CI 0.66-0.91) and women (P = 0.02; OR 0.83; 95% CI 0.71-0.97) after adjusting for systemic risk factors using multivariable logistic regression. There was no significant impact of sex on the relationship between AL and DR (P = 0.56). In the subset of patients with asymmetric DR, the percentage of patients whose shorter eye had a higher stage of DR was not significantly different between men and women (P = 0.20). Male sex is a risk factor for DR in a diverse safety-net hospital population. Longer AL is associated with a lower risk of DR, and this relationship is not affected by sex.
| S-EPMC9192741 | biostudies-literature
Project description:To determine the time and risk factors for developing proliferative diabetic retinopathy (PDR) and vitreous hemorrhage (VH).Multicenter, national cohort study.Anonymized data of 50 254 patient eyes with diabetes mellitus at 19 UK hospital eye services were extracted at the initial and follow-up visits between 2007 and 2014. Time to progression of PDR and VH were calculated with Cox regression after stratifying by baseline diabetic retinopathy (DR) severity and adjusting for age, sex, race, and starting visual acuity.Progression to PDR in 5 years differed by baseline DR: no DR (2.2%), mild (13.0%), moderate (27.2%), severe nonproliferative diabetic retinopathy (NPDR) (45.5%). Similarly, 5-year progression to VH varied by baseline DR: no DR (1.1%), mild (2.9%), moderate (7.3%), severe NPDR (9.8%). Compared with no DR, the patient eyes that presented with mild, moderate, and severe NPDR were 6.71, 14.80, and 28.19 times more likely to develop PDR, respectively. In comparison to no DR, the eyes with mild, moderate, and severe NPDR were 2.56, 5.60, and 7.29 times more likely to develop VH, respectively. In severe NPDR, the eyes with intraretinal microvascular abnormalities (IRMA) had a significantly increased hazard ratio (HR) of developing PDR (HR 1.77, 95% confidence interval [CI] 1.25-2.49, P = .0013) compared with those with venous beading, whereas those with 4-quadrant dot-blot hemorrhages (4Q DBH) had 3.84 higher HR of developing VH (95% CI 1.39-10.62, P = .0095).Baseline severities and features of initial DR are prognostic for PDR development. IRMA increases risk of PDR whereas 4Q DBH increases risk of VH.
| S-EPMC5608549 | biostudies-literature
Project description:ObjectiveDespite recent calls for home healthcare (HHC) to integrate informatics, the application of machine learning in HHC is relatively unknown. Thus, this study aimed to synthesize and appraise the literature describing the application of machine learning to predict adverse outcomes (e.g., hospitalization, mortality) using electronic health record (EHR) data in the HHC setting. Our secondary aim was to evaluate the comprehensiveness of predictors used in the machine learning algorithms guided by the Biopsychosocial Model.MethodsDuring March 2022 we conducted a literature search in four databases: PubMed, Embase, CINAHL, and Scopus. Inclusion criteria were 1) describing services provided in the HHC setting, 2) applying machine learning algorithms to predict adverse outcomes, defined as outcomes related to patient deterioration, 3) using EHR data and 4) focusing on the adult population. Predictors were mapped to the Biopsychosocial Model. A risk of bias analysis was conducted using the Prediction Model Risk Of Bias Assessment Tool.ResultsThe final sample included 20 studies. Eighteen studies used predictors from standardized assessments integrated in the EHR. The most common outcome of interest was hospitalization (55%), followed by mortality (25%). Psychological predictors were frequently excluded (35%). Tree based algorithms were most frequently applied (75%). Most studies demonstrated high or unclear risk of bias (75%).ConclusionFuture studies in HHC should consider incorporating machine learning algorithms into clinical decision support systems to identify patients at risk. Based on the Biopsychosocial model, psychological and interpersonal characteristics should be used along with biological characteristics to enhance risk prediction. To facilitate the widespread adoption of machine learning, stakeholders should encourage standardization in the HHC setting.
| S-EPMC9869861 | biostudies-literature
Project description:ContextDiabetic retinopathy (DR) is the leading cause of incident blindness among working-age adults in the United States. Federally designated safety-net clinics (FDSC) often serve as point-of-contact for patients least likely to receive recommended DR screenings, creating opportunity for targeted interventions to increase screening access and compliance.Study design and methodsWith such a goal, we implemented and assessed the longitudinal performance of an FDSC-based telemedicine DR screening (TDRS) network of 22 clinical sites providing nonmydriatic fundus photography with remote interpretation and reporting. Retrospective analysis of patient encounters between February 2014 and January 2019 was performed to assess rates of pathology and referral. A generalized estimating equation logistic regression model was used for subset analysis from audits of pre- and post-implementation screening rates. Finally, patient surveys were conducted and assessed as a measure of intervention acceptability.ResultsOf the 13,923 individual telescreening encounters (4327 female, 4220 male, and 5376 unspecified; mean [SD] age, 54.9 [12.5] years) studied, 10,540 were of adequate quality to identify 3532 (33.5%) patients with ocular pathology: 2319 (22.0%) patients had some level of DR with 1604 (15.2%) requiring specialist referral, and 808 (7.7%) patients required referral for other ocular pathologies. The mean screening rate for audited clinics in the year prior to program implementation was 29.9% (641/2147), which increased to 47.7% (1012/2124) in the program's first year, doubling patients' odds of being screened (OR 2.2; 95% CI: 1.3-3.7; P = .003). These gains were sustained over four years following implementation (OR 1.9; 95% CI: 1.1-3.1; P = .018) despite varied clinic screening performance (4-year averaged range, 22.9-55.1%). Odds of early detection likewise doubled for patients with consecutive screenings (OR 2.2, 95% CI: 2.0-2.4; P < .001). Finally, surveyed patients preferred TDRS to specialist exams (82.5%; 776/941) and would recommend the service to friends (92.7%; 868/936).Conclusion and relevanceA statewide, FDSC-centered TDRS network was successfully established and sustained in a medically underserved region of the United States. Our results suggest that large TDRS networks in FDSCs can increase screening access and compliance for otherwise unscreened populations, but outcomes can vary greatly among clinics. Further work to optimize program implementation is needed to maximize this model's impact.
| S-EPMC7641408 | biostudies-literature
Project description:IntroductionPrevention of tuberculosis disease through diagnosis and treatment of latent tuberculosis infection is critical for achieving tuberculosis elimination in the U.S. Diagnosis and treatment of latent tuberculosis infection in safety-net primary care settings that serve patients at risk for tuberculosis may increase uptake of this prevention effort and accelerate progress toward elimination. Optimizing tuberculosis prevention in these settings requires measuring the latent tuberculosis infection care cascade (testing, diagnosis, and treatment) and identifying gaps to develop solutions to overcome barriers. We used electronic health record data to describe the latent tuberculosis infection care cascade and identify gaps among a network of safety-net primary care clinics.MethodsElectronic health record data for patients seen in the OCHIN Clinical Network, the largest network of safety-net clinics in the U.S., between 2012 and 2019 were extracted. electronic health record data were used to measure the latent tuberculosis infection care cascade: patients who met tuberculosis screening criteria on the basis of current recommendations were tested for tuberculosis infection, diagnosed with latent tuberculosis infection, and prescribed treatment for latent tuberculosis infection. Outcomes were stratified by diagnostic test and treatment regimen.ResultsAmong 1.9 million patients in the analytic cohort, 43.5% met tuberculosis screening criteria, but only 21.4% were tested for latent tuberculosis infection; less than half (40.4%) were tested using an interferon-gamma release assay. Among those with a valid result, 10.5% were diagnosed with latent tuberculosis infection, 29.1% of those were prescribed latent tuberculosis infection treatment, and only 33.6% were prescribed a recommended rifamycin-based regimen.ConclusionsElectronic health record data can be used to measure the latent tuberculosis infection care cascade. A large proportion of patients in this safety-net clinical network are at high risk for tuberculosis infection. Addressing identified gaps in latent tuberculosis infection testing and treatment may have a direct impact on improving tuberculosis prevention in primary care clinics and accelerate progress toward elimination.
| S-EPMC10630620 | biostudies-literature
Project description:AimElectronic healthcare record (EHR)-based surveillance systems are increasingly being developed to support early detection of safety signals. It is unknown what the power of such a system is for surveillance among children and adolescents. In this paper we provide estimates of the number and classes of drugs, and incidence rates (IRs) of events, that can be monitored in children and adolescents (0-18 years).MethodsData were obtained from seven population-based EHR databases in Denmark, Italy, and the Netherlands during the period 1996-2010. We estimated the number of drugs for which specific adverse events can be monitored as a function of actual drug use, minimally detectable relative risk (RR) and IRs for 10 events.ResultsThe population comprised 4 838 146 individuals (25 575 132 person years (PYs)), who were prescribed 2170 drugs (1 610 631 PYs drug-exposure). Half of the total drug-exposure in PYs was covered by only 18 drugs (0.8%). For a relatively frequent event like upper gastrointestinal bleeding there were 39 drugs for which an association with a RR ≥4, if present, could be investigated. The corresponding number of drugs was eight for a rare event like anaphylactic shock.ConclusionDrug use in children is rare and shows little variation. The number of drugs with enough exposure to detect rare adverse events in children and adolescents within an EHR-based surveillance system such as EU-ADR is limited. Use of additional sources of paediatric drug exposure information and global collaboration are imperative in order to optimize EHR data for paediatric safety surveillance.
| S-EPMC4541978 | biostudies-literature
Project description:BackgroundMost hospitals use traditional infection prevention (IP) methods for outbreak detection. We developed the Enhanced Detection System for Healthcare-Associated Transmission (EDS-HAT), which combines whole-genome sequencing (WGS) surveillance and machine learning (ML) of the electronic health record (EHR) to identify undetected outbreaks and the responsible transmission routes, respectively.MethodsWe performed WGS surveillance of healthcare-associated bacterial pathogens from November 2016 to November 2018. EHR ML was used to identify the transmission routes for WGS-detected outbreaks, which were investigated by an IP expert. Potential infections prevented were estimated and compared with traditional IP practice during the same period.ResultsOf 3165 isolates, there were 2752 unique patient isolates in 99 clusters involving 297 (10.8%) patient isolates identified by WGS; clusters ranged from 2-14 patients. At least 1 transmission route was detected for 65.7% of clusters. During the same time, traditional IP investigation prompted WGS for 15 suspected outbreaks involving 133 patients, for which transmission events were identified for 5 (3.8%). If EDS-HAT had been running in real time, 25-63 transmissions could have been prevented. EDS-HAT was found to be cost-saving and more effective than traditional IP practice, with overall savings of $192 408-$692 532.ConclusionsEDS-HAT detected multiple outbreaks not identified using traditional IP methods, correctly identified the transmission routes for most outbreaks, and would save the hospital substantial costs. Traditional IP practice misidentified outbreaks for which transmission did not occur. WGS surveillance combined with EHR ML has the potential to save costs and enhance patient safety.
| S-EPMC9427134 | biostudies-literature
Project description:BackgroundArtificial intelligence and automated technology were first reported more than 70 years ago and nowadays provide unprecedented diagnostic accuracy, screening capacity, risk stratification, and workflow optimization. Diabetic retinopathy is an important cause of preventable blindness worldwide, and artificial intelligence technology provides precocious diagnosis, monitoring, and guide treatment. High-quality exams are fundamental in supervised artificial intelligence algorithms, but the lack of ground truth standards in retinal exams datasets is a problem.Main bodyIn this article, ETDRS, NHS, ICDR, SDGS diabetic retinopathy grading, and manual annotation are described and compared in publicly available datasets. The various DR labeling systems generate a fundamental problem for AI datasets. Possible solutions are standardization of DR classification and direct retinal-finding identifications.ConclusionReliable labeling methods also need to be considered in datasets with more trustworthy labeling.
| S-EPMC8722080 | biostudies-literature
Project description:BackgroundThe rapid expansion in the use of electronic health records (EHR) has increased the number of medical errors originating in health information systems (HIS). The sociotechnical approach helps in understanding risks in the development, implementation, and use of EHR and health information technology (HIT) while accounting for complex interactions of technology within the health care system.ObjectiveThis study addresses two important questions: (1) "which of the common EHR error types are associated with perceived high- and extreme-risk severity ratings among EHR users?", and (2) "which variables are associated with high- and extreme-risk severity ratings?"MethodsThis study was a quantitative, non-experimental, descriptive study of EHR users. We conducted a cross-sectional web-based questionnaire study at the largest hospital district in Finland. Statistical tests included the reliability of the summative scales tested with Cronbach's alpha. Logistic regression served to assess the association of the independent variables to each of the eight risk factors examined.ResultsA total of 2864 eligible respondents provided the final data. Almost half of the respondents reported a high level of risk related to the error type "extended EHR unavailability". The lowest overall risk level was associated with "selecting incorrectly from a list of items". In multivariate analyses, profession and clinical unit proved to be the strongest predictors for high perceived risk. Physicians perceived risk levels to be the highest ( P<.001 in six of eight error types), while emergency departments, operating rooms, and procedure units were associated with higher perceived risk levels ( P<.001 in four of eight error types). Previous participation in eLearning courses on EHR-use was associated with lower risk for some of the risk factors.ConclusionsBased on a large number of Finnish EHR users in hospitals, this study indicates that HIT safety hazards should be taken very seriously, particularly in operating rooms, procedure units, emergency departments, and intensive care units/critical care units. Health care organizations should use proactive and systematic assessments of EHR risks before harmful events occur. An EHR training program should be compulsory for all EHR users in order to address EHR safety concerns resulting from the failure to use HIT appropriately.
| S-EPMC4890731 | biostudies-literature
Project description:ObjectiveTo study the relation between electronic health record (EHR) variables and healthcare process events.Materials and methodsLagged linear correlation was calculated between five healthcare process events and 84 EHR variables (24 clinical laboratory values and 60 clinical concepts extracted from clinical notes) in a 24-year database. The EHR variables were clustered for each healthcare process event and interpreted.ResultsLaboratory tests tended to cluster together and note concepts tended to cluster together. Within each of those two classes, the variables clustered into clinically sensible groupings. The exact groupings varied from healthcare process event to event, with the largest differences occurring between inpatient events and outpatient events.DiscussionUnlike previously reported pairwise associations between variables, which highlighted correlations across the laboratory-clinical note divide, incorporating healthcare process events appeared to be sensitive to the manner in which the variables were collected.ConclusionWe believe that it may be possible to exploit this sensitivity to help knowledge engineers select variables and correct for biases.
| S-EPMC3861922 | biostudies-literature