Background: Automated classification of glomerular pathological findings is potentially beneficial in establishing an efficient and objective diagnosis in renal pathology. While previous studies have verified the artificial intelligence (AI) models for the classification of global sclerosis and glomerular cell proliferation, there are several other glomerular pathological findings required for diagnosis, and the comprehensive models for the classification of these major findings have not yet been reported. Whether the cooperation between these AI models and clinicians improves diagnostic performance also remains unknown. Here, we developed AI models to classify glomerular images for major findings required for pathological diagnosis and investigated whether those models could improve the diagnostic performance of nephrologists. Methods: We used a dataset of 283 kidney biopsy cases comprising 15,888 glomerular images that were annotated by a total of 25 nephrologists. AI models to classify seven pathological findings: global sclerosis, segmental sclerosis, endocapillary proliferation, mesangial matrix accumulation, mesangial cell proliferation, crescent, and basement membrane structural changes, were constructed using deep learning by fine-tuning of InceptionV3 convolutional neural network. Subsequently, we compared the agreement to truth labels between majority decision among nephrologists with or without the AI model as a voter. Results: Our model for global sclerosis showed high performance (area under the curve: periodic acid-Schiff, 0.986; periodic acid methenamine silver, 0.983); the models for the other findings also showed performance close to those of nephrologists. By adding the AI model output to majority decision among nephrologists, out of the 14 constructed models, the results of the majority decision showed improvement in sensitivity for 10 models (four of them were statistically significant) and specificity for eight models (five significant). Conclusion: Our study showed a proof-of-concept for the classification of multiple glomerular findings in a comprehensive method of deep learning and suggested its potential effectiveness in improving diagnostic accuracy of clinicians.

ObjectiveTo investigate the cost-effectiveness of screening and subsequent intervention for age-related macular degeneration (AMD) in Japan.MethodsThe clinical effectiveness and cost-effectiveness of screening and subsequent intervention for AMD were assessed using a Markov model. The Markov model simulation began at the age of 40 years and concluded at the age of 90 years. The first-eye and second-eye combined model assumed an annual state-transition probability, development of prodromal symptoms, choroidal neovascularization (CNV), and reduction in visual acuity. Anti–vascular-endothelial-growth-factor (anti-VEGF) intravitreal injection therapy and photodynamic therapy (PDT) were performed to treat CNV. Intake of supplements was recommended to patients who had prodromal symptoms and unilateral AMD. Data on prevalence, morbidity, transition probability, utility value of each AMD patient, and treatment costs were obtained from published clinical reports.ResultsIn the base-case analysis, screening for AMD every 5 years, beginning at the age of 50 years, showed a decrease of 41% in the total number of blind patients. The screening program reduced the incidence of blindness more than did the additional intake of supplements. However, the incremental cost-effectiveness ratio (ICER) of screening versus no screening was 27,486,352 Japanese yen (JPY), or 259,942 US dollars (USD) per quality-adjusted life year (QALY). In the sensitivity analysis, prodromal symptom-related factors for AMD had great impacts on the cost-effectiveness of screening. The lowest ICER obtained from the best scenario was 4,913,717 JPY (46,470 USD) per QALY, which was approximately equal to the willingness to pay in Japan.ConclusionsOphthalmologic screening for AMD in adults is highly effective in reducing the number of patients with blindness but not cost-effective as demonstrated by a Markov model based on clinical data from Japan.

BackgroundAcetaminophen (APAP) is frequently used for analgesia and is considered safer than nonsteroidal anti-inflammatory drugs (NSAIDs) for the kidneys. However, there is little epidemiological evidence of the association between APAP and acute kidney injury (AKI).ObjectivesTo examine the association between APAP and AKI using the self-controlled case series (SCCS) method, which is a novel strategy to control between-person confounders by comparing the risk and reference periods in each patient.MethodsWe performed SCCS in 1,871 patients (39.9% female) who were administered APAP and subsequently developed AKI, by reviewing electronically stored hospital information system data from May 2011 to July 2016. We used conditional Poisson regression to compare each patient’s risk and reference period. As a time-varying confounder, we adjusted the status of liver and kidney functions, systemic inflammation, and exposure to NSAIDs.ResultsWe identified 5,650 AKI events during the 260,549 person-day observation period. The unadjusted incidences during the reference and exposure periods were 2.01/100 and 3.12/100 person-days, respectively. The incidence rate ratio adjusted with SCCS was 1.03 (95% confidence interval [CI]: 0.95–1.12). When we restricted endpoints as stage 2 AKI- and stage 3 AKI-level creatinine elevations, the incidence rate ratios were 1.20 (95% CI 0.91–1.58) and 1.20 (95% CI 0.62–2.31), respectively, neither of which was statistically significant.ConclusionOur findings added epidemiological information for the relationship between APAP administration and AKI development. The results indicated scarce association between APAP and AKI, presumably supporting the general physicians’ impression that APAP is safer for kidney.

AimsThe latest evidence in the incidence of central retinal artery occlusion (CRAO) is needed to support the development of novel treatments as orphan drugs. However, up-to-date information on the incidence of CRAO in the ageing or aged population is limited. We aimed to investigate the nationwide epidemiological and clinical characteristics of CRAO in Japan, using nationwide health insurance claims data.MethodsWe analysed a total of 16 069 762 claims data in the sampling dataset of the National Database of Health Insurance Claims and Specific Health Checkups of Japan (NDB), which is the nationwide health insurance claims database of 127 million whole Japanese individuals. CRAO was identified using the International Classification of Diseases 10th edition diagnostic code H34.1. The crude incidence rates and age-standardised incidence rates of CRAO, according to the standard age-structure population of the WHO, were calculated.ResultsThe crude incidence rate of CRAO in Japan was 5.84 (95% CI, 5.71 to 5.97) per 100 000 person-years. With respect to the sex-related incidence, the rate was higher 1.40 times in men than in women (6.85 (95% CI, 6.65 to 7.06) vs 4.88 (95% CI, 4.71 to 5.05), p<0.001). The age-standardised incidence rate was 2.53 (95% CI, 2.29 to 2.76) per 100 000 person-years.ConclusionsThe crude incidence rate of CRAO was higher in Japan than in other countries, as reported previously, reflecting the Japanese population structure as a super-aged society. These findings can be helpful for the development of appropriate healthcare policies to address the increasing incidence of CRAO with the ageing population.

Background: Automated classification of glomerular pathological findings is potentially beneficial in establishing an efficient and objective diagnosis in renal pathology. While previous studies have verified the artificial intelligence (AI) models for the classification of global sclerosis and glomerular cell proliferation, there are several other glomerular pathological findings required for diagnosis, and the comprehensive models for the classification of these major findings have not yet been reported. Whether the cooperation between these AI models and clinicians improves diagnostic performance also remains unknown. Here, we developed AI models to classify glomerular images for major findings required for pathological diagnosis and investigated whether those models could improve the diagnostic performance of nephrologists. Methods: We used a dataset of 283 kidney biopsy cases comprising 15888 glomerular images that were annotated by a total of 25 nephrologists. AI models to classify seven pathological findings: global sclerosis, segmental sclerosis, endocapillary proliferation, mesangial matrix accumulation, mesangial cell proliferation, crescent, and basement membrane structural changes, were constructed using deep learning by fine-tuning of InceptionV3 convolutional neural network. Subsequently, we compared the agreement to truth labels between majority decision among nephrologists with or without the AI model as a voter. Results: Our model for global sclerosis showed high performance (area under the curve: periodic acid-Schiff, 0.986; periodic acid methenamine silver, 0.983); the models for the other findings also showed performance close to those of nephrologists. By adding the AI model output to majority decision among nephrologists, the sensitivity and specificity were significantly improved in 9 of 14 constructed models compared to those of nephrologists alone. Conclusion: Our study showed a proof-of-concept for the classification of multiple glomerular findings in a comprehensive method of deep learning and suggested its potential effectiveness in improving diagnostic accuracy of clinicians.