Self-Supervised Learning for Time Series Analysis: Taxonomy, Progress, and Prospects.

Self-supervised learning (SSL) has recently achieved impressive performance on various time series tasks. The most prominent advantage of SSL is that it reduces the dependence on labeled data. Based on the pre-training and fine-tuning strategy, even a small amount of labeled data can achieve high performance. Compared with many published self-supervised surveys on computer vision and natural language processing, a comprehensive survey for time series SSL is still missing. To fill this gap, we review current state-of-the-art SSL methods for time series data in this article. To this end, we first comprehensively review existing surveys related to SSL and time series, and then provide a new taxonomy of existing time series SSL methods by summarizing them from three perspectives: generative-based, contrastive-based, and adversarial-based. These methods are further divided into ten subcategories with detailed reviews and discussions about their key intuitions, main frameworks, advantages and disadvantages. To facilitate the experiments and validation of time series SSL methods, we also summarize datasets commonly used in time series forecasting, classification, anomaly detection, and clustering tasks. Finally, we present the future directions of SSL for time series analysis.

Mutational spectrum for guiding the decision of adjuvant treatment in patients with resected biliary tract carcinoma.

BACKGROUND: Systemic chemotherapy or chemoradiation therapy has proven to be effective in treating advanced biliary tract carcinoma (BTC). However, its efficacy in the adjuvant setting remains controversial. Therefore, this study aimed to determine the prognostic significance of genomic biomarkers in resected BTC and their potential role in stratifying patients for adjuvant treatment. METHODS: We retrospectively reviewed 113 BTC patients who underwent curative-intent surgery and had available tumor sequencing data. Disease-free survival (DFS) was the primary outcome examined and univariate analysis was used to identify gene mutations with prognostic value. Favorable and unfavoratble gene subsets were distinguished from the selected genes through grouping, respectively. Multivariate Cox regression was used to identify independent prognostic factors of DFS. RESULTS: Our results indicated that mutations in ACVR1B, AR, CTNNB1, ERBB3, and LRP2 were favorable mutations, while mutations in ARID1A, CDKN2A, FGFR2, NF1, NF2, PBRM1, PIK3CA, and TGFBR1 were unfavorable mutations. In addition to age, sex, and node positive, favorable genes (HR = 0.15, 95% CI = 0.04-0.48, p = 0.001) and unfavorable genes (HR = 2.86, 95% CI = 1.51-5.29, p = 0.001) were identified as independent prognostic factors for DFS. Out of the 113 patients, only 35 received adjuvant treatment whereas the majority (78) did not. For patients with both favorable and unfavorable mutations undetected, adjuvant treatment showed negative effect on DFS (median DFS: S441 vs. 956 days, p = 0.010), but there was no significant difference in DFS among those in other mutational subgroups. CONCLUSIONS: Genomic testing might be useful in guiding the decisions regarding adjuvant treatment in BTC.

microRNA-329 suppresses epithelial-to-mesenchymal transition and lymph node metastasis in bile duct cancer by inhibiting laminin subunit beta 3.

Bile duct cancer (BDC), also known as cholangiocarcinoma, is a highly desmoplastic cancer with a growth pattern characterized by periductal extension and infiltration. Studies have suggested that microRNAs (miRNAs) play an important role in BDC progression. Here we aim at investigating the effects of miR-329 on BDC development, focusing especially on epithelial-to-mesenchymal transition (EMT) in vitro and lymph node metastasis in vivo. Expression microarrays associated with BDC tissues were collected and differentially expressed genes were analyzed, followed by miRNA target prediction and verification. The role miR-329 played in BDC was examined using gain-of-function and loss-of-function methods. The expressions of miR-329, laminin subunit beta 3 (LAMB3), and EMT markers, in addition to cell proliferation, migration, and invasion were evaluated. Furthermore, nude mice models of BDC were established to observe tumor growth and metastatic lymph nodes. The LAMB3 was identified as an upregulated gene based on the GSE77984 and GSE45001 microarray analysis. LAMB3 was also predicted and confirmed to be a target gene of miR-329 by dual-luciferase reporter assay. Through further cell experiments, the EMT process was reversed, cell proliferation, invasion, and migration were suppressed, when miR-329 was upregulated. Furthermore, in vivo experiments exhibited that the overexpression of miR-329 inhibited tumor growth and the number of metastatic lymph nodes. This study provides in vivo and in vitro evidence that miR-329 inhibits BDC progression through translational repression of LAMB3. Therefore, the obtained results may aid as an experimental basis for improving prognosis of BDC.

Outcomes of non-anatomic liver resection for hepatocellular carcinoma in the patients with liver cirrhosis and analysis of prognostic factors.

PURPOSE: In the east countries, patients with hepatocellular carcinoma (HCC) are usually associated with varied degrees of liver cirrhosis, and anatomic resection is therefore limited to use, especially in those with severe liver cirrhosis. This study aims to evaluate the clinical value of non-anatomic resection in HCC patients with cirrhosis. METHODS: Seventy-seven consecutive HCC patients with cirrhosis underwent non-anatomic liver resection in Tongji Hospital from January 2003 to December 2006. The clinical data, severity of liver cirrhosis, and survival rates of these patients were retrospectively evaluated, and the prognostic factors were analyzed. RESULTS: One-, 2-, and 3-year overall and disease-free survival rates of this cohort of patients were 78%, 68%, 56%, and 66%, 58%, 55%, respectively. The hospital mortality and morbidity were 0% and 24.7%, respectively. The 1-, 2-, and 3-year overall survival rates were 85.7%, 77.1%, and 74.3% in the patients with mild cirrhosis, 81.5%, 63%, and 48.1% in the patients with moderate cirrhosis, and 60.0%, 53.3%, and 26.7% in the patients with severe cirrhosis, respectively. There was a significant difference among the patients with different grades of cirrhosis (P = 0.001). Multivariate and univariate analyses revealed that severity of cirrhosis, tumor diameter larger than 5 cm, and vascular invasion were independent prognostic factors. CONCLUSIONS: Non-anatomic liver resection for HCC could yield comparable outcomes with anatomic resection in the patients with mild cirrhosis or tumors diameter smaller than 5 cm. Severity of cirrhosis is an independent factor worsening long-time survival. Non-anatomic resection is a safe and effective surgical modality in the treatment of HCC patients with cirrhosis.