While healthcare data sets include extensive information about patients, many researchers have limitations in analyzing them due to their intrinsic characteristics such as heterogeneity, longitudinal irregularity, and noise. In particular, since the majority of medical history information is recorded in text codes, the use of such information has been limited due to the high dimensionality of explanatory variables. To address this problem, recent studies applied word embedding techniques, originally developed for natural language processing, and derived positive results in terms of dimensional reduction and accuracy of the prediction model. This paper reviews the deep learning-based natural language processing techniques (word embedding) and summarizes research cases that have used those techniques in the health care field. Then we finally propose a research framework for applying deep learning-based natural language process in the analysis of domestic health insurance data.

Background: and Objective: With the advancement of cancer treatments and increased life expectancy, managing breakthrough cancer pain (BTcP) is essential to improve the quality of life for cancer patients. This study aimed to compare the major rapid onset opioids in Korea based on their characteristics and costs to determine the best option for each patient. Methods: Based on sales information from IQVIA-MIDAS, sublingual fentanyl tablet (SLF), fentanyl buccal tablet (FBT), and oral transmucosal fentanyl citrate (OTFC) were selected as the top three drugs for the treatment of BTcP in Korea, considering them the most comparable drugs. The cost and cost-pain relief ratio of the drugs for short-term (1 month) and long-term (1 year) treatment were compared and the ease of administration based on various factors, including pharmacokinetics, onset of action, and administration procedures were evaluated. Results: SLF was evaluated as the best overall in terms of rapid onset of action, ease of administration, and drug cost and also had the highest market share. SLF had the lowest cost pain relief ratio for both the initial and supplemental treatment for the 1-month pain intensity difference 15 (PID15) ratio. However, for the 1-month PID30 ratio, SLF was not superior to OTFC or FBT. The longer the breakthrough cancer pain duration, the more cost-effective the other rapid onset opioids. Conclusion The rapid onset opioids that fit the patient’s breakthrough cancer pain pattern have the best cost-effectiveness.

Background: Highly pathogenic avian influenza viruses (HPAIVs) is an extremely contagious and high mortality rates in chickens resulting in substantial economic impact on the poultry sector. Therefore, it is necessary to elucidate the pathogenic mechanism of HPAIV for infection control. Objective: Gene set enrichment analysis (GSEA) can effectively avoid the limitations of subjective screening for differential gene expression. Therefore, we performed GSEA to compare HPAI-infected resistant and susceptible Ri chicken lines. Methods: The Ri chickens Mx(A)/BF2(B21) were chosen as resistant, and the chickens Mx(G)/ BF2(B13) were selected as susceptible by genotyping the Mx and BF2 genes. The tracheal tissues of HPAIV H5N1 infected chickens were collected for RNA sequencing followed by GSEA analysis to define gene subsets to elucidate the sequencing results. Results: We identified four differentially expressed pathways, which were immune-related pathways with a total of 78 genes. The expression levels of cytokines (IL-1β, IL-6, IL-12), chemokines (CCL4 and CCL5), type interferons and their receptors (IFN-β, IFNAR1, IFNAR2, and IFNGR1), Jak-STAT signaling pathway genes (STAT1, STAT2, and JAK1), MHC class I and II and their co-stimulatory molecules (CD80, CD86, CD40, DMB2, BLB2, and B2M), and interferon stimulated genes (EIF2AK2 and EIF2AK1) in resistant chickens were higher than those in susceptible chickens. Conclusions Resistant Ri chickens exhibit a stronger antiviral response to HPAIV H5N1 compared with susceptible chickens. Our findings provide insights into the immune responses of genetically disparate chickens against HPAIV.

Mutations in the Leucine-rich repeat kinase 2 (LRRK2 ) gene are the most prevalent cause of familial Parkinson’s disease (PD). The increase in LRRK2 kinase activity observed in the pathogenic G2019S mutation is important for PD development. Several studies have reported that increased LRRK2 kinase activity and treatment with LRRK2 kinase inhibitors decreased and increased ciliogenesis, respectively, in mouse embryonic fibroblasts (MEFs) and retinal pigment epithelium (RPE) cells. In contrast, treatment of SH-SY5Y dopaminergic neuronal cells with PD-causing chemicals increased ciliogenesis. Because these reports were somewhat contradictory, we tested the effect of LRRK2 kinase activity on ciliogenesis in neurons. In SH-SY5Y cells, LRRK2 inhibitor treatment slightly increased ciliogenesis, but serum starvation showed no increase. In rat primary neurons, LRRK2 inhibitor treatment repeatedly showed no significant change. Little difference was observed between primary cortical neurons prepared from wild-type (WT) and G2019S +/- mice. However, a significant increase in ciliogenesis was observed in G2019S +/- compared to WT human fibroblasts, and this pattern was maintained in neural stem cells (NSCs) differentiated from the induced pluripotent stem cells (iPSCs) prepared from the same WT/G2019S fibroblast pair. NSCs differentiated from G2019S and its gene-corrected WT counterpart iPSCs were also used to test ciliogenesis in an isogenic background. The results showed no significant difference between WT and G2019S regardless of kinase inhibitor treatment and B27-deprivation-mimicking serum starvation. These results suggest that LRRK2 kinase activity may be not a direct regulator of ciliogenesis and ciliogenesis varies depending upon the cell type or genetic background.

Mutations in the Leucine-rich repeat kinase 2 (LRRK2 ) gene are the most prevalent cause of familial Parkinson’s disease (PD). The increase in LRRK2 kinase activity observed in the pathogenic G2019S mutation is important for PD development. Several studies have reported that increased LRRK2 kinase activity and treatment with LRRK2 kinase inhibitors decreased and increased ciliogenesis, respectively, in mouse embryonic fibroblasts (MEFs) and retinal pigment epithelium (RPE) cells. In contrast, treatment of SH-SY5Y dopaminergic neuronal cells with PD-causing chemicals increased ciliogenesis. Because these reports were somewhat contradictory, we tested the effect of LRRK2 kinase activity on ciliogenesis in neurons. In SH-SY5Y cells, LRRK2 inhibitor treatment slightly increased ciliogenesis, but serum starvation showed no increase. In rat primary neurons, LRRK2 inhibitor treatment repeatedly showed no significant change. Little difference was observed between primary cortical neurons prepared from wild-type (WT) and G2019S +/- mice. However, a significant increase in ciliogenesis was observed in G2019S +/- compared to WT human fibroblasts, and this pattern was maintained in neural stem cells (NSCs) differentiated from the induced pluripotent stem cells (iPSCs) prepared from the same WT/G2019S fibroblast pair. NSCs differentiated from G2019S and its gene-corrected WT counterpart iPSCs were also used to test ciliogenesis in an isogenic background. The results showed no significant difference between WT and G2019S regardless of kinase inhibitor treatment and B27-deprivation-mimicking serum starvation. These results suggest that LRRK2 kinase activity may be not a direct regulator of ciliogenesis and ciliogenesis varies depending upon the cell type or genetic background.