No abstract

Semi-supervised learning and continuous learning are fundamental paradigms for human-level intelligence. To deal with real-world problems where labels are rarely given and the opportunity to access the same data is limited, it is necessary to apply these two paradigms in a joined fashion. In this paper, we propose Label Propagation Adaptive Resonance Theory (LPART) for semi-supervised continuous learning. LPART uses an online label propagation mechanism to perform classification and gradually improves its accuracy as the observed data accumulates. We evaluated the proposed model on visual (MNIST, SVHN, CIFAR-10) and audio (NSynth) datasets by adjusting the ratio of the labeled and unlabeled data. The accuracies are much higher when both labeled and unlabeled data are used, demonstrating the significant advantage of LPART in environments where the data labels are scarce.

Background: Although emerging evidence suggest acute kidney injury (AKI) progress to chronic kidney disease (CKD), long-term renal outcome of AKI still remains unclear. Acute tubular necrosis (ATN) is the most common cause of AKI due to ischemia, toxin or sepsis. Acute interstitial nephritis (AIN), caused by drugs or autoimmune diseases is also increasingly recognized as an important cause of AKI. Unlike glomerular diseases, AKI is usually diagnosed in the clinical context without kidney biopsies, and lack of histology might contribute to this uncertainty. Methods: Among 8,769 biopsy series, 253 adults who were histologically diagnosed with ATN and AIN from 1982 to 2018 at five university hospitals were included. Demographic and pathological features that are associated with the development of end stage renal disease (ESRD) were also examined. Results: Rate of non-recovery of renal function at 6 month was significantly higher in the AIN (ATN vs AIN 49.3 vs 69.4%, P = 0.007) with a 2.71-fold higher risk of non-recovery compared to ATN (95% confidence interval [CI], 1.20-6.47). During the mean follow up of 76.5 ± 91.9 months, ESRD developed in 39.4% of patients with AIN, and 21.5% patients of ATN. The risk of ESRD was significantly higher in AIN (23.05; 95% CI, 2.42-219.53) and also in ATN (12.14; 95% CI, 1.19-24.24) compared to control with non-specific pathology. Older age, female gender, renal function at the time of biopsy and at 6 months, proteinuria and pathological features including interstitial inflammation and fibrosis, tubulitis, vascular lesion were significantly associated with progression to ESRD. Conclusion: Our study demonstrated that patients with biopsy proven ATN and AIN are at high risk of developing ESRD. AIN showed higher rate of non-renal recovery at 6 month than ATN.

No abstract

No abstract