An innovative supervised longitudinal learning procedure of recurrent neural networks with temporal data augmentation: Insights from predicting fetal macrosomia and large-for-gestational age.

BACKGROUND: Longitudinal data in health informatics studies often present challenges due to sparse observations from each subject, limiting the application of contemporary deep learning for prediction. This issue is particularly relevant in predicting birthweight, a crucial factor in identifying conditions such as macrosomia and large-for-gestational age (LGA). Previous approaches have relied on empirical formulas for estimated fetal weights (EFWs) from ultrasound measurements and mixed-effects models for interim predictions. METHOD: The proposed novel supervised longitudinal learning procedure features a three-step approach. First, EFWs are generated using empirical formulas from ultrasound measurements. Second, nonlinear mixed-effects models are applied to create augmented sequences of EFWs, spanning daily gestational timepoints. This augmentation transforms sparse longitudinal data into a dense parallel sequence suitable for training recurrent neural networks (RNNs). A tailored RNN architecture is then devised to incorporate the augmented sequential EFWs along with non-sequential maternal characteristics. RESULTS: The RNNs are trained on augmented data to predict birthweights, which are further classified for macrosomia and LGA. Application of this supervised longitudinal learning procedure to the Successive Small-for-Gestational-Age Births study yields improved performance in classification metrics. Specifically, sensitivity, area under the receiver operation characteristic curve, and Youden's Index demonstrate enhanced results, indicating the effectiveness of the proposed approach in overcoming sparsity challenges in longitudinal health informatics data. CONCLUSIONS: The integration of mixed-effects models for temporal data augmentation and RNNs on augmented sequences shows effective in accurately predicting birthweights, particularly in the context of identifying excessive fetal growth conditions.

Epigenetic silencing of BEND4, a novel DNA damage repair gene, is a synthetic lethal marker for ATM inhibitor in pancreatic cancer.

Synthetic lethality is a novel model for cancer therapy. To understand the function and mechanism of BEN domain-containing protein 4 (BEND4) in pancreatic cancer, eight cell lines and a total of 492 cases of pancreatic neoplasia samples were included in this study. Methylation-specific polymerase chain reaction, CRISPR/Cas9, immunoprecipitation assay, comet assay, and xenograft mouse model were used. BEND4 is a new member of the BEN domain family. The expression of BEND4 is regulated by promoter region methylation. It is methylated in 58.1% (176/303) of pancreatic ductal adenocarcinoma (PDAC), 33.3% (14/42) of intraductal papillary mucinous neoplasm, 31.0% (13/42) of pancreatic neuroendocrine tumor, 14.3% (3/21) of mucinous cystic neoplasm, 4.3% (2/47) of solid pseudopapillary neoplasm, and 2.7% (1/37) of serous cystic neoplasm. BEND4 methylation is significantly associated with late-onset PDAC (> 50 years, P < 0.01) and tumor differentiation (P < 0.0001), and methylation of BEND4 is an independent poor prognostic marker (P < 0.01) in PDAC. Furthermore, BEND4 plays tumor-suppressive roles in vitro and in vivo. Mechanistically, BEND4 involves non-homologous end joining signaling by interacting with Ku80 and promotes DNA damage repair. Loss of BEND4 increased the sensitivity of PDAC cells to ATM inhibitor. Collectively, the present study revealed an uncharacterized tumor suppressor BEND4 and indicated that methylation of BEND4 may serve as a potential synthetic lethal marker for ATM inhibitor in PDAC treatment.

Methylation of NRIP3 Is a Synthetic Lethal Marker for Combined PI3K and ATR/ATM Inhibitors in Colorectal Cancer.

INTRODUCTION: The aim of this study was to investigate the epigenetic regulation and underlying mechanism of NRIP3 in colorectal cancer (CRC). METHODS: Eight cell lines (SW480, SW620, DKO, LOVO, HT29, HCT116, DLD1, and RKO), 187 resected margin samples from colorectal cancer tissue, 146 cases with colorectal adenomatous polyps, and 308 colorectal cancer samples were used. Methylation-specific PCR, Western blotting, RNA interference assay, and a xenograft mouse model were used. RESULTS: NRIP3 exhibited methylation in 2.7% (5/187) of resected margin samples from colorectal cancer tissue, 32.2% (47/146) of colorectal adenomatous polyps, and 50.6% (156/308) of CRC samples, and the expression of NRIP3 was regulated by promoter region methylation. The methylation of NRIP3 was found to be significantly associated with late onset (at age 50 years or older), poor tumor differentiation, lymph node metastasis, and poor 5-year overall survival in CRC (all P < 0.05). In addition, NRIP3 methylation was an independent poor prognostic marker ( P < 0.05). NRIP3 inhibited cell proliferation, colony formation, invasion, and migration, while induced G1/S arrest. NRIP3 suppressed CRC growth by inhibiting PI3K-AKT signaling both in vitro and in vivo . Methylation of NRIP3 sensitized CRC cells to combined PI3K and ATR/ATM inhibitors. DISCUSSION: NRIP3 was frequently methylated in both colorectal adenomatous polyps and CRC. The methylation of NRIP3 may potentially serve as an early detection, late-onset, and poor prognostic marker in CRC. NRIP3 is a potential tumor suppressor. NRIP3 methylation is a potential synthetic lethal marker for combined PI3K and ATR/ATM inhibitors.