Ocular biomarkers: useful incidental findings by deep learning algorithms in fundus photographs.

BACKGROUND/OBJECTIVES: Artificial intelligence can assist with ocular image analysis for screening and diagnosis, but it is not yet capable of autonomous full-spectrum screening. Hypothetically, false-positive results may have unrealized screening potential arising from signals persisting despite training and/or ambiguous signals such as from biomarker overlap or high comorbidity. The study aimed to explore the potential to detect clinically useful incidental ocular biomarkers by screening fundus photographs of hypertensive adults using diabetic deep learning algorithms. SUBJECTS/METHODS: Patients referred for treatment-resistant hypertension were imaged at a hospital unit in Perth, Australia, between 2016 and 2022. The same 45° colour fundus photograph selected for each of the 433 participants imaged was processed by three deep learning algorithms. Two expert retinal specialists graded all false-positive results for diabetic retinopathy in non-diabetic participants. RESULTS: Of the 29 non-diabetic participants misclassified as positive for diabetic retinopathy, 28 (97%) had clinically useful retinal biomarkers. The models designed to screen for fewer diseases captured more incidental disease. All three algorithms showed a positive correlation between severity of hypertensive retinopathy and misclassified diabetic retinopathy. CONCLUSIONS: The results suggest that diabetic deep learning models may be responsive to hypertensive and other clinically useful retinal biomarkers within an at-risk, hypertensive cohort. Observing that models trained for fewer diseases captured more incidental pathology increases confidence in signalling hypotheses aligned with using self-supervised learning to develop autonomous comprehensive screening. Meanwhile, non-referable and false-positive outputs of other deep learning screening models could be explored for immediate clinical use in other populations.

Locally derived traffic-related air pollution and fetal growth restriction: a retrospective cohort study.

BACKGROUND: Fetal growth restriction has been inconsistently associated with maternal exposure to elevated levels of traffic-related air pollution. OBJECTIVE: We investigated the relationship between an individualised measure of fetal growth and maternal exposure to a specific marker for traffic-related air pollution. METHODS: We estimated maternal residential exposure to a marker for traffic-related air pollution (nitrogen dioxide, NO2) during pregnancy for 23,452 births using temporally adjusted land-use regression. Logistic regression was used to investigate associations with small for gestational age and sex (SGA) and fetal growth restriction, defined as proportion of optimal birth weight (POBW) below the 10th percentile. Sub-populations investigated were: women who spent most time at home, women who did not move house, women with respiratory or circulatory morbidity, women living in low/middle/high socio-economic areas, women who delivered before 37 weeks gestation, and women who delivered from 37 weeks gestation. RESULTS: An IQR increase in traffic-related air pollution in the second trimester across all women was associated with an OR of 1.31 (95% CI 1.07 to 1.60) for fetal growth restriction. Effects on fetal growth restriction (low POBW) were highest among women who subsequently delivered before 37 weeks of gestation. Effects on SGA were highest among women who did not move house: OR 1.35 (95% CI 1.08 to 1.69). CONCLUSIONS: Larger effect sizes were observed for low POBW than for SGA. Exposure to traffic-related air pollution in mid to late pregnancy was associated with risk of SGA and low POBW in this study.

Use of a total traffic count metric to investigate the impact of roadways on asthma severity: a case-control study.

BACKGROUND: This study had two principal objectives: (i) to investigate the relationship between asthma severity and proximity to major roadways in Perth, Western Australia; (ii) to demonstrate a more accurate method of exposure assessment for traffic pollutants using an innovative GIS-based measure that fully integrates all traffic densities around subject residences. METHODS: We conducted a spatial case-control study, in which 'cases' were defined as individuals aged under 19 years of age with more severe asthma (defined here as two or more emergency department contacts with asthma in a defined 5-year period) versus age- and gender-matched 'controls' with less severe asthma (defined here as one emergency department contact for asthma). Traffic exposures were measured using a GIS-based approach to determine the lengths of the roads falling within a buffer area, and then multiplying them by their respective traffic counts. RESULTS: We examined the spatial relationship between emergency department contacts for asthma at three different buffer sizes: 50 metres, 100 metres and 150 metres. No effect was noted for the 50 metre buffer (OR = 1.07; 95% CI: 0.91-1.26), but elevated odds ratios were observed with for crude (unadjusted) estimates OR = 1.21 (95% CI: 1.00-1.46) for 100 metre buffers and OR = 1.25 (95% CI: 1.02-1.54) for 150 metre buffers. For adjusted risk estimates, only the 150 metre buffer yielded a statistically significant finding (OR = 1.24; 95% CI:1.00-1.52). CONCLUSIONS: Our study revealed a significant 24% increase in the risk of experiencing multiple emergency department contacts for asthma for every log-unit of traffic exposure. This study provides support for the hypothesis that traffic related air pollution increases the frequency of health service contacts for asthma. This study used advanced GIS techniques to establish traffic-weighted buffer zones around the geocoded residential location of subjects to provide an accurate assessment of exposure to traffic emissions, thereby providing a quantification of the ranges over which pollutants may exert a health effect.

An insight into the relationships between hepcidin, anemia, infections and inflammatory cytokines in pediatric refugees: a cross-sectional study.

BACKGROUND: Hepcidin, a key regulator of iron homeostasis, is increased in response to inflammation and some infections, but the in vivo role of hepcidin, particularly in children with iron deficiency anemia (IDA) is unclear. We investigated the relationships between hepcidin, cytokines and iron status in a pediatric population with a high prevalence of both anemia and co-morbid infections. METHODOLOGY/PRINCIPAL FINDINGS: African refugee children <16 years were consecutively recruited at the initial post-resettlement health check with 181 children meeting inclusion criteria. Data on hematological parameters, cytokine levels and co-morbid infections (Helicobacter pylori, helminth and malaria) were obtained and urinary hepcidin assays performed. The primary outcome measure was urinary hepcidin levels in children with and without iron deficiency (ID) and/or ID anaemia (IDA). The secondary outcome measures included were the relationship between co-morbid infections and (i) ID and IDA, (ii) urinary hepcidin levels and (iii) cytokine levels. IDA was present in 25/181 (13.8%). Children with IDA had significantly lower hepcidin levels (IDA median hepcidin 0.14 nmol/mmol Cr (interquartile range 0.05-0.061) versus non-IDA 2.96 nmol/mmol Cr, (IQR 0.95-6.72), p<0.001). Hemoglobin, log-ferritin, iron, mean cell volume (MCV) and transferrin saturation were positively associated with log-hepcidin levels (log-ferritin beta coefficient (beta): 1.30, 95% CI 1.02 to 1.57) and transferrin was inversely associated (beta: -0.12, 95% CI -0.15 to -0.08). Cytokine levels (including IL-6) and co-morbid infections were not associated with IDA or hepcidin levels. CONCLUSIONS/SIGNIFICANCE: This is the largest pediatric study of the in vivo associations between hepcidin, iron status and cytokines. Gastro-intestinal infections (H. pylori and helminths) did not elevate urinary hepcidin or IL-6 levels in refugee children, nor were they associated with IDA. Longitudinal and mechanistic studies of IDA will further elucidate the role of hepcidin in paediatric iron regulation.

Diagnosis of Helicobacter pylori infection in a high-prevalence pediatric population: a comparison of 2 fecal antigen testing methods and serology.

OBJECTIVES: Accurate methods for diagnosing active Helicobacter pylori infection in children have been limited to invasive or time-consuming techniques. Recently, fecal antigen testing has been used successfully for the diagnosis of H pylori infection in the pediatric population. We compared 2 monoclonal fecal antigen diagnostic methods in a population of children with a suspected high prevalence of H pylori infection. We also assessed the diagnostic performance of H pylori immunoglobulin G serology. MATERIALS AND METHODS: In a cross-sectional study of African refugee children (<16 years) we compared an immunochromatographic technique (ICT) and serology with a monoclonal fecal antigen enzyme immunoassay (MFAT) method for the detection of active H pylori infection. Following the manufacturer's instructions, an optical density of >or=0.190 was used as a cutoff for MFAT. Sensitivity, specificity, and positive and negative predictive values were calculated. RESULTS: Of the 193 eligible children enrolled, active H pylori infection was detected in 149 of 182 (81.9%) in whom MFAT was performed. The prevalence of active infection increased with age; children with active infection were significantly older, and there were no sex differences. ICT and serology underperformed in comparison with MFAT (ICT sensitivity 74.6%, specificity 63.6%, positive predictive value 89.8%, negative predictive value 36.8%; and serology sensitivity 57.9%, specificity 77.4%, positive predictive value 92.0%, negative predictive value 29.9%). CONCLUSIONS: Monoclonal enzyme immunoassay fecal antigen testing is a practical and feasible alternative to traditional invasive diagnostic methods in high-prevalence pediatric populations. Neither immunochromatography nor serology is useful for the diagnosis of active H pylori infection in these children.