Browsing by Author "Li, Xin"
- Results Per Page
- Sort Options
Item Open Access A COMPARISON OF MARKET CLEARING TOOLS IN ELECTRICITY SYSTEMS(2015-04-23) Li, XinTo maintain a strict balance between demand and supply in the US power systems, the Independent System Operators (ISOs) schedule power plants and determine electricity prices using a market clearing model. This model determines for each time period and power plant, the times of startup, shutdown, the amount of power production, and the provisioning of spinning and non-spinning power generation reserves, etc. Such a deterministic optimization model takes as input the characteristics of all the generating units such as their power generation installed capacity, ramp rates, minimum up and down time requirements, and marginal costs for production, as well as the forecast of intermittent energy such as wind and solar, along with the minimum reserve requirement of the whole system. This reserve requirement is determined based on the likelihood of outages on the supply side and on the levels of error forecasts in demand and intermittent generation. With increased installed capacity of intermittent renewable energy, determining the appropriate level of reserve requirements has become harder. Stochastic market clearing models have been proposed as an alternative to deterministic market clearing models. Rather than using a fixed reserve targets as an input, stochastic market clearing models take different scenarios of wind power into consideration and determine reserves schedule as output. Using a scaled version of the power generation system of PJM, a regional transmission organization (RTO) that coordinates the movement of wholesale electricity in all or parts of 13 states and the District of Columbia, and wind scenarios generated from BPA (Bonneville Power Administration) data, this paper explores a comparison of the performance between a stochastic and deterministic model in market clearing. The two models are compared in their ability to contribute to the affordability, reliability and sustainability of the electricity system, measured in terms of total operational costs, load shedding and air emissions. The process of building the models and running for tests indicate that a fair comparison is difficult to obtain due to the multi-dimensional performance metrics considered here, and the difficulty in setting up the parameters of the models in a way that does not advantage or disadvantage one modeling framework. Along these lines, this study explores the effect that model assumptions such as reserve requirements, value of lost load (VOLL) and wind spillage costs have on the comparison of the performance of stochastic vs deterministic market clearing models.Item Open Access Advancing Deep-Generated Speech and Defending against Its Misuse(2023) Cai, ZexinDeep learning has revolutionized speech generation, spanning synthesis areas such as text-to-speech and voice conversion, leading to diverse advancements. On the one hand, when trained on high-quality datasets, artificial voices now exhibit a level of synthesized quality that rivals human speech in naturalness. On the other, cutting-edge deep synthesis research is making strides in producing controllable systems, allowing for generating audio signals in arbitrary voice and speaking style.
Yet, despite their impressive synthesis capabilities, current speech generation systems still face challenges in controlling and manipulating speech attributes. Control over crucial attributes, such as speaker identity and language, essential for enhancing the functionality of a synthesis system, still needs to be improved. Specifically, systems capable of cloning a target speaker's voice in cross-lingual contexts or replicating unseen voices are still in their nascent stages. On the other hand, the heightened naturalness of synthesized speech has raised concerns, posing security threats to both humans and automated speech processing systems. The rise of accessible audio deepfakes, capable of spreading misinformation or bypassing biometric security, accentuates the complex interplay between advancing and defencing against deep-synthesized speech.
Consequently, this dissertation delves into the dynamics of deep-generated speech, viewing it from two perspectives. Offensively, we aim to enhance synthesis systems to elevate their capabilities. On the defensive side, we introduce methodologies to counter emerging audio deepfake threats, offering solutions grounded in detection-based approaches and reliable synthesis system design.
Our research yields several noteworthy findings and conclusions. First, we present an improved voice cloning method incorporated with our novel feedback speaker consistency mechanism. Second, we demonstrate the feasibility of achieving cross-lingual multi-speaker speech synthesis with a limited amount of bilingual data, offering a synthesis method capable of producing diverse audio across various speakers and languages. Third, our proposed frame-level detection model for partially fake audio attacks proves effective in detecting tampered utterances and locating the modified regions within. Lastly, by employing an invertible synthesis system, we can trace back to the original speaker of a converted utterance. Despite these strides, each domain of our study still confronts challenges, further fueling our motivation for persistent research and refinement of the associated performance.
Item Open Access Applying Machine Learning to Testing and Diagnosis of Integrated Systems(2021) Pan, RenjianThe growing complexity of integrated boards and systems makes manufacturing test and diagnosis increasingly expensive. There is a pressing need to reduce test cost and to pinpoint the root causes of integrated systems in a more effective way. In light of machine learning, a number of intelligent test-cost reduction and root-cause analysis methods have been proposed. However, it remains extremely challenging to (i) reduce test cost for black-box testing for integrated systems, and (ii) pinpoint the root causes for integrated systems with little need on labeled test data from repair history. To tackle these challenges, we propose multiple machine-learning-based solutions for black-box test-cost reduction and unsupervised/semi-supervised root-cause analysis in this dissertation.For black-box test-cost reduction, we propose a novel test selection method based on a Bayesian network model. First, it is formulated as a constrained optimization problem. Next, a score-based algorithm is implemented to construct the Bayesian network for black-box tests. Finally, we propose a Bayesian index with the property of Markov blankets, and then an iterative test selection method is developed based on our proposed Bayesian index. For root-cause analysis, we first propose an unsupervised root-cause analysis method in which no repair history is needed. In the first stage, a decision-tree model is trained with system test information to cluster the data in a coarse-grained manner. In the second stage, frequent-pattern mining is applied to extract frequent patterns in each decision-tree node to precisely cluster the data so that each cluster represents only a small number of root causes. The proposed method can accommodate both numerical and categorical test items. A combination of the L-method, cross validation and Silhouette score enables us to automatically determine all hyper-parameters. Two industry case studies with system test data demonstrate that the proposed approach significantly outperforms the state-of-the-art unsupervised root-cause-analysis method. Utilizing transfer learning, we further improve the performance of unsupervised root-cause-analysis. A two-stage clustering method is first developed by exploiting model selection based on the concept of Silhouette score. Next, a data-selection method based on ensemble learning is proposed to transfer valuable information from a source product to improve the diagnosis accuracy on the target product with insufficient data. Two case studies based on industry designs demonstrate that the proposed approach significantly outperforms other state-of-the-art unsupervised root-cause-analysis methods. In addition, we propose a semi-supervised root-cause-analysis method with co-training, where only a small set of labeled data is required. Using random forest as the learning kernel, a co-training technique is proposed to leverage the unlabeled data by automatically pre-labeling a subset of them and retraining each decision tree. In addition, several novel techniques have been proposed to avoid over-fitting and determine hyper-parameters. Two case studies based on industrial designs demonstrate that the proposed approach significantly outperforms the state-of-the-art methods. In summary, this dissertation addresses the most difficult problems in testing and diagnosis of integrated systems with machine learning. A test selection method based on Bayesian networks reduces the test cost for black-box testing. With unsupervised learning, semi-supervised learning and transfer learning, we analysis root causes for integrated systems without much need on historical diagnosis information. The proposed approaches are expected to contribute to the semiconductor industry by effectively reducing the black-box test cost and efficiently diagnosing the integrated systems.
Item Open Access Association of genetic variants of TMEM135 and PEX5 in the peroxisome pathway with cutaneous melanoma-specific survival.(Annals of translational medicine, 2021-03) Wang, Haijiao; Liu, Hongliang; Dai, Wei; Luo, Sheng; Amos, Christopher I; Lee, Jeffrey E; Li, Xin; Yue, Ying; Nan, Hongmei; Wei, QingyiBackground
Peroxisomes are ubiquitous and dynamic organelles that are involved in the metabolism of reactive oxygen species (ROS) and lipids. However, whether genetic variants in the peroxisome pathway genes are associated with survival in patients with melanoma has not been established. Therefore, our aim was to identify additional genetic variants in the peroxisome pathway that may provide new prognostic biomarkers for cutaneous melanoma (CM).Methods
We assessed the associations between 8,397 common single-nucleotide polymorphisms (SNPs) in 88 peroxisome pathway genes and CM disease-specific survival (CMSS) in a two-stage analysis. For the discovery, we extracted the data from a published genome-wide association study from The University of Texas MD Anderson Cancer Center (MDACC). We then replicated the results in another dataset from the Nurse Health Study (NHS)/Health Professionals Follow-up Study (HPFS).Results
Overall, 95 (11.1%) patients in the MDACC dataset and 48 (11.7%) patients in the NHS/HPFS dataset died of CM. We found 27 significant SNPs in the peroxisome pathway genes to be associated with CMSS in both datasets after multiple comparison correction using the Bayesian false-discovery probability method. In stepwise Cox proportional hazards regression analysis, with adjustment for other covariates and previously published SNPs in the MDACC dataset, we identified 2 independent SNPs (TMEM135 rs567403 C>G and PEX5 rs7969508 A>G) that predicted CMSS (P=0.003 and 0.031, respectively, in an additive genetic model). The expression quantitative trait loci analysis further revealed that the TMEM135 rs567403 GG and PEX5 rs7969508 GG genotypes were associated with increased and decreased levels of mRNA expression of their genes, respectively.Conclusions
Once our findings are replicated by other investigators, these genetic variants may serve as novel biomarkers for the prediction of survival in patients with CM.Item Open Access Efficacy and Safety of a Pegasparaginase-Based Chemotherapy Regimen vs an L-asparaginase-Based Chemotherapy Regimen for Newly Diagnosed Advanced Extranodal Natural Killer/T-Cell Lymphoma: A Randomized Clinical Trial.(JAMA oncology, 2022-06-16) Wang, Xinhua; Zhang, Lei; Liu, Xiangli; Li, Xin; Li, Ling; Fu, Xiaorui; Sun, Zhenchang; Wu, Jingjing; Zhang, Xudong; Yan, Jiaqin; Chang, Yu; Nan, Feifei; Zhou, Zhiyuan; Wu, Xiaolong; Tian, Li; Ma, Minrui; Li, Zhaoming; Yu, Hui; Zhu, Linan; Wang, Yingjun; Shi, Cunzhen; Feng, Xiaoyan; Li, Jiwei; Ding, Mengjie; Zhang, Jieming; Dong, Meng; Xue, Hongwei; Wang, Jinghua; Zou, Liqun; Su, Liping; Wu, Jianqiu; Liu, Lihong; Bao, Huizheng; Zhang, Liling; Guo, Yanzhen; Guo, Shuxia; Lu, Yi; Young, Ken H; Li, Wencai; Zhang, MingzhiImportance
The L-asparaginase-based SMILE (dexamethasone, methotrexate, ifosfamide, L-asparaginase, and etoposide) chemotherapy regimen has shown higher response rates and survival benefit over an anthracycline-containing regimen. However, the safety profile was not satisfied. A well-tolerated regimen with promising efficacy is lacking.Objective
To compare the efficacy and safety of the DDGP (dexamethasone, cisplatin, gemcitabine, and pegaspargase) regimen with the SMILE regimen in newly diagnosed advanced-stage (III/IV) extranodal natural killer/T-cell lymphoma (ENKL).Design, setting, and participants
This was an open-label, multicenter, randomized clinical trial that took place across 12 participating hospitals in China from January 2011 to February 2019. Patients were eligible if they were 14 to 70 years old with newly diagnosed ENKL in stages III/IV and had an Eastern Cooperative Oncology Group performance status of 0 to 2. Eligible patients were evenly randomized to either the DDGP or SMILE group.Interventions
Patients in each group were treated with the assigned regimen every 21 days for 6 cycles.Main outcomes and measures
The primary end point was progression-free survival (PFS), and secondary end points included overall response rate and overall survival (OS). The adverse events between the DDGP and SMILE groups were compared.Results
Among the 87 randomized patients, 80 received treatment (40 in the DDGP group and 40 in the SMILE group); the median (IQR) age was 43 (12) years, and 51 (64%) were male. The baseline characteristics were similar between the groups. At a median follow-up of 41.5 months, the median PFS was not reached in the DDGP group vs 6.8 months in the SMILE group (HR, 0.42; 95% CI, 0.23-0.77; P = .004), and the median OS was not reached in the DDGP group vs 75.2 months in the SMILE group (HR, 0.41; 95% CI, 0.19-0.89, P = .02). The PFS rate at 3 years and OS rate at 5 years were higher in the DDGP group vs the SMILE group (3-year PFS, 56.6% vs 41.8%; 5-year OS, 74.3% vs 51.7%). The overall response rate was higher in the DDGP group than in the SMILE group (90.0% vs 60.0%; P = .002). Grade 3 and 4 hematologic toxic effects were more frequently reported in the SMILE group vs the DDGP group (leukopenia, 85.0% vs 62.5%; neutropenia, 85.0% vs 65.0%).Conclusions and relevance
In this randomized clinical trial, the DDGP regimen showed promising preliminary results for patients with newly diagnosed local advanced ENKL. A confirmation trial based on larger population is warranted.Trial registration
ClinicalTrials.gov Identifier: NCT01501149.Item Open Access Genetic variants in ELOVL2 and HSD17B12 predict melanoma-specific survival.(International journal of cancer, 2019-02-08) Dai, Wei; Liu, Hongliang; Xu, Xinyuan; Ge, Jie; Luo, Sheng; Zhu, Dakai; Amos, Christopher I; Fang, Shenying; Lee, Jeffrey E; Li, Xin; Nan, Hongmei; Li, Chunying; Wei, QingyiFatty acids play a key role in cellular bioenergetics, membrane biosynthesis and intracellular signaling processes and thus may be involved in cancer development and progression. In the present study, we comprehensively assessed associations of 14,522 common single-nucleotide polymorphisms (SNPs) in 149 genes of the fatty-acid synthesis pathway with cutaneous melanoma disease-specific survival (CMSS). The dataset of 858 cutaneous melanoma (CM) patients from a published genome-wide association study (GWAS) by The University of Texas M.D. Anderson Cancer Center was used as the discovery dataset, and the identified significant SNPs were validated by a dataset of 409 CM patients from another GWAS from the Nurses' Health and Health Professionals Follow-up Studies. We found 40 noteworthy SNPs to be associated with CMSS in both discovery and validation datasets after multiple comparison correction by the false positive report probability method, because more than 85% of the SNPs were imputed. By performing functional prediction, linkage disequilibrium analysis, and stepwise Cox regression selection, we identified two independent SNPs of ELOVL2 rs3734398 T>C and HSD17B12 rs11037684 A>G that predicted CMSS, with an allelic hazards ratio of 0.66 (95% confidence interval = 0.51-0.84 and p = 8.34 × 10-4 ) and 2.29 (1.55-3.39 and p = 3.61 × 10-5 ), respectively. Finally, the ELOVL2 rs3734398 variant CC genotype was found to be associated with a significantly increased mRNA expression level. These SNPs may be potential markers for CM prognosis, if validated by additional larger and mechanistic studies.Item Open Access Genetic variants in PDSS1 and SLC16A6 of the ketone body metabolic pathway predict cutaneous melanoma-specific survival.(Molecular carcinogenesis, 2020-06) Dai, Wei; Liu, Hongliang; Chen, Ka; Xu, Xinyuan; Qian, Danwen; Luo, Sheng; Amos, Christopher I; Lee, Jeffrey E; Li, Xin; Nan, Hongmei; Li, Chunying; Wei, QingyiA few single-nucleotide polymorphisms (SNPs) have been identified to be associated with cutaneous melanoma (CM) survival through genome-wide association studies, but stringent multiple testing corrections required for the hypothesis-free testing may have masked some true associations. Using a hypothesis-driven analysis approach, we sought to evaluate associations between SNPs in ketone body metabolic pathway genes and CM survival. We comprehensively assessed associations between 4196 (538 genotyped and 3658 imputed) common SNPs in 44 ketone body metabolic pathway genes and CM survival, using a dataset of 858 patients of a case-control study from The University of Texas M.D. Anderson Cancer Center as the discovery set and another dataset of 409 patients from the Nurses' Health Study and the Health Professionals Follow-up Study as the replication set. There were 95/858 (11.1%) and 48/409 (11.7%) patients who died of CM, respectively. We identified two independent SNPs (ie, PDSS1 rs12254548 G>C and SLC16A6 rs71387392 G>A) that were associated with CM survival, with allelic hazards ratios of 0.58 (95% confidence interval [CI] = 0.44-0.76, P = 9.00 × 10-5 ) and 1.98 (95% CI = 1.34-2.94, P = 6.30 × 10-4 ), respectively. Additionally, associations between genotypes of the SNPs and messenger RNA expression levels of their corresponding genes support the biologic plausibility of a role for these two variants in CM tumor progression and survival. Once validated by other larger studies, PDSS1 rs12254548 and SLC16A6 rs71387392 may be valuable biomarkers for CM survival.Item Open Access Genetic variants in RORA and DNMT1 associated with cutaneous melanoma survival.(International journal of cancer, 2018-06) Li, Bo; Wang, Yanru; Xu, Yinghui; Liu, Hongliang; Bloomer, Wendy; Zhu, Dakai; Amos, Christopher I; Fang, Shenying; Lee, Jeffrey E; Li, Xin; Han, Jiali; Wei, QingyiCutaneous melanoma (CM) is considered as a steroid hormone-related malignancy. However, few studies have evaluated the roles of genetic variants encoding steroid hormone receptor genes and their related regulators (SHR-related genes) in CM-specific survival (CMSS). Here, we performed a pathway-based analysis to evaluate genetic variants of 191 SHR-related genes in 858 CMSS patients using a dataset from a genome-wide association study (GWAS) from The University of Texas MD Anderson Cancer Center (MDACC), and then validated the results in an additional dataset of 409 patients from the Harvard GWAS. Using multivariate Cox proportional hazards regression analysis, we identified three-independent SNPs (RORA rs782917 G > A, RORA rs17204952 C > T and DNMT1 rs7253062 G > A) as predictors of CMSS, with a variant-allele attributed hazards ratio (HR) and 95% confidence interval of 1.62 (1.25-2.09), 1.60 (1.20-2.13) and 1.52 (1.20-1.94), respectively. Combined analysis of risk genotypes of these three SNPs revealed a decreased CMSS in a dose-response manner as the number of risk genotypes increased (ptrend < 0.001); however, no improvement in the prediction model was observed (area under the curve [AUC] = 79.6-80.8%, p = 0.656), when these risk genotypes were added to the model containing clinical variables. Our findings suggest that genetic variants of RORA and DNMT1 may be promising biomarkers for CMSS, but these results needed to be validated in future larger studies.Item Open Access Genetic variants in the calcium signaling pathway genes are associated with cutaneous melanoma-specific survival.(Carcinogenesis, 2018-12-29) Wang, Xiaomeng; Liu, Hongliang; Xu, Yinghui; Xie, Jichun; Zhu, Dakai; Amos, Christopher I; Fang, Shenying; Lee, Jeffrey E; Li, Xin; Nan, Hongmei; Song, Yanqiu; Wei, QingyiRemodeling or deregulation of the calcium signaling pathway is a relevant hallmark of cancer including cutaneous melanoma (CM). In the present study, using data from a published genome-wide association study (GWAS) from The University of Texas M.D. Anderson Cancer Center, we assessed the role of 41,377 common single nucleotide polymorphisms (SNPs) of 167 calcium signaling pathway genes in CM survival. We used another GWAS from Harvard University as the validation dataset. In the single-locus analysis, 1,830 SNPs were found to be significantly associated with CM-specific survival (CMSS) (P ≤ 0.050 and false-positive report probability ≤ 0.2), of which nine SNPs were validated in the Harvard study (P ≤ 0.050). Among these, three independent SNPs (i.e., PDE1A rs6750552 T>C, ITPR1 rs6785564 A>G and RYR3 rs2596191 C>A) had a predictive role in CMSS, with a meta-analysis derived hazards ratio (HR) of 1.52 [95% confidence interval (CI) = 1.19-1.94, P = 7.21×10-4]], 0.49 (0.33-0.73, 3.94×10-4) and 0.67 (0.53-0.86, 0.0017), respectively. Patients with an increasing number of protective genotypes had remarkably improved CMSS. Additional expression quantitative trait loci (eQTL) analysis showed that these genotypes were also significantly associated with mRNA expression levels of the genes. Taken together, these results may help us to identify prospective biomarkers in the calcium signaling pathway for CM prognosis.Item Open Access Genetic variants in the metzincin metallopeptidase family genes predict melanoma survival.(Molecular carcinogenesis, 2018-01) Xu, Yinghui; Wang, Yanru; Liu, Hongliang; Shi, Qiong; Zhu, Dakai; Amos, Christopher I; Fang, Shenying; Lee, Jeffrey E; Hyslop, Terry; Li, Xin; Han, Jiali; Wei, QingyiMetzincins are key molecules in the degradation of the extracellular matrix and play an important role in cellular processes such as cell migration, adhesion, and cell fusion of malignant tumors, including cutaneous melanoma (CM). We hypothesized that genetic variants of the metzincin metallopeptidase family genes would be associated with CM-specific survival (CMSS). To test this hypothesis, we first performed Cox proportional hazards regression analysis to evaluate the associations between genetic variants of 75 metzincin metallopeptidase family genes and CMSS using the dataset from the genome-wide association study (GWAS) from The University of Texas MD Anderson Cancer Center (MDACC) which included 858 non-Hispanic white patients with CM, and then validated using the dataset from the Harvard GWAS study which had 409 non-Hispanic white patients with invasive CM. Four independent SNPs (MMP16 rs10090371 C>A, ADAMTS3 rs788935 T>C, TLL2 rs10882807 T>C and MMP9 rs3918251 A>G) were identified as predictors of CMSS, with a variant-allele attributed hazards ratio (HR) of 1.73 (1.32-2.29, 9.68E-05), 1.46 (1.15-1.85, 0.002), 1.68 (1.31-2.14, 3.32E-05) and 0.67 (0.51-0.87, 0.003), respectively, in the meta-analysis of these two GWAS studies. Combined analysis of risk genotypes of these four SNPs revealed a decreased CMSS in a dose-response manner as the number of risk genotypes increased (Ptrend < 0.001). An improvement was observed in the prediction model (area under the curve [AUC] = 81.4% vs. 78.6%), when these risk genotypes were added to the model containing non-genotyping variables. Our findings suggest that these genetic variants may be promising prognostic biomarkers for CMSS.Item Open Access Genetic variants in TKT and DERA in the nicotinamide adenine dinucleotide phosphate pathway predict melanoma survival.(European journal of cancer (Oxford, England : 1990), 2020-07-09) Gu, Ning; Dai, Wei; Liu, Hongliang; Ge, Jie; Luo, Sheng; Cho, Eunyoung; Amos, Christopher I; Lee, Jeffrey E; Li, Xin; Nan, Hongmei; Yuan, Hua; Wei, QingyiBACKGROUND:Cutaneous melanoma (CM) is the most lethal type of skin cancers. Nicotinamide adenine dinucleotide phosphate (NADPH) plays an important role in anabolic reactions and tumorigenesis, but many genes are involved in the NADPH system. METHODS:We used 10,912 single-nucleotide polymorphisms (SNPs) (2018 genotyped and 8894 imputed) in 134 NADPH-related genes from a genome-wide association study (GWAS) of 858 patients from The University of Texas MD Anderson Cancer Center (MDACC) in a single-locus analysis to predict CM survival. We then replicated the results in another GWAS data set of 409 patients from the Nurses' Health Study (NHS) and the Health Professionals Follow-up Study (HPFS). RESULTS:There were 95 of 858 (11.1%) and 48 of 409 (11.7%) patients who died of CM, respectively. In multivariable Cox regression analyses, we identified two independent SNPs (TKT rs9864057 G > A and deoxyribose phosphate aldolase (DERA) rs12297652 A > G) to be significantly associated with CM-specific survival [hazards ratio (HR) of 1.52, 95% confidence interval (CI) = 1.18-1.96, P = 1.06 × 10-3 and 1.51 (1.19-1.91, 5.89 × 10-4)] in the meta-analysis, respectively. Furthermore, an increasing number of risk genotypes of these two SNPs was associated with a higher risk of death in the MDACC, the NHS/HPFS, and their combined data sets (Ptrend<0.001, = 0.004 and <0.001, respectively). In the expression quantitative trait loci analysis, TKT rs9864057 G > A and DERA rs12297652 A > G were also significantly associated with higher mRNA expression levels in sun-exposed lower-leg skin (P = 0.043 and 0.006, respectively). CONCLUSIONS:These results suggest that these two potentially functional SNPs may be valuable prognostic biomarkers for CM survival, but larger studies are needed to validate these findings.Item Open Access Genetic variants of SDCCAG8 and MAGI2 in mitosis-related pathway genes are independent predictors of cutaneous melanoma-specific survival.(Cancer science, 2021-08-10) He, Yuanmin; Liu, Hongliang; Luo, Sheng; Amos, Christopher I; Lee, Jeffrey E; Li, Xin; Nan, Hongmei; Wei, QingyiMitosis is a prognostic factor for cutaneous melanoma (CM), but accurate mitosis detection in CM tissues is difficult. Therefore, the 8th Edition of the American Joint Committee on Cancer staging system has removed mitotic rate as a category criterion of the tumor T-category, based on the evidence that mitotic rate was not an independent prognostic factor for melanoma survival. Since single-nucleotide polymorphisms (SNPs) have been shown to be potential predictors for cutaneous melanoma-specific survival (CMSS), we investigated the potential prognostic value of SNPs in mitosis-related pathway genes in CMSS by analyzing their associations with outcomes of 850 CM patients from The University of Texas MD Anderson Cancer Center in a discovery dataset and validated the findings in another dataset of 409 CM patients from the Harvard University Nurses' Health Study and Health Professionals Follow-up Study. In both datasets, we identified two SNPs (SDCCAG8 rs10803138 G>A and MAGI2 rs3807694 C>T) as independent prognostic factors for CMSS, with adjusted allelic hazards ratios of 1.49 (95% confidence interval=1.17-1.90, P=0.001) and 1.45 (1.13-1.86, P=0.003), respectively. Furthermore, their combined unfavorable alleles also predicted poor survival in both discovery and validation datasets in a dose-response manner (Ptrend =0.0006 and 0.0001, respectively). Additional functional analysis revealed that both SDCCAG8 rs10803138 A and MAGI2 rs3807694 T alleles were associated with elevated mRNA expression levels in normal tissues. Therefore, these findings suggest that SDCCAG8 rs10803138 G>A and MAGI2 rs3807694 C>T are independent prognostic biomarkers for CMSS, possibly by regulating the mRNA expression of the corresponding genes involved in mitosis.Item Open Access Intrahepatic microbes govern liver immunity by programming NKT cells(Journal of Clinical Investigation, 2022-04-15) Leinwand, Joshua C; Paul, Bidisha; Chen, Ruonan; Xu, Fangxi; Sierra, Maria A; Paluru, Madan M; Nanduri, Sumant; Alcantara, Carolina G; Shadaloey, Sorin AA; Yang, Fan; Adam, Salma A; Li, Qianhao; Bandel, Michelle; Gakhal, Inderdeep; Appiah, Lara; Guo, Yuqi; Vardhan, Mridula; Flaminio, Zia; Grodman, Emilie R; Mermelstein, Ari; Wang, Wei; Diskin, Brian; Aykut, Berk; Khan, Mohammad; Werba, Gregor; Pushalkar, Smruti; McKinstry, Mia; Kluger, Zachary; Park, Jaimie J; Hsieh, Brandon; Dancel-Manning, Kristen; Liang, Feng-Xia; Park, James S; Saxena, Anjana; Li, Xin; Theise, Neil D; Saxena, Deepak; Miller, GeorgeItem Open Access Novel Genetic Variants of ALG6 and GALNTL4 of the Glycosylation Pathway Predict Cutaneous Melanoma-Specific Survival.(Cancers, 2020-01-24) Zhou, Bingrong; Zhao, Yu Chen; Liu, Hongliang; Luo, Sheng; Amos, Christopher I; Lee, Jeffrey E; Li, Xin; Nan, Hongmei; Wei, QingyiBecause aberrant glycosylation is known to play a role in the progression of melanoma, we hypothesize that genetic variants of glycosylation pathway genes are associated with the survival of cutaneous melanoma (CM) patients. To test this hypothesis, we used a Cox proportional hazards regression model in a single-locus analysis to evaluate associations between 34,096 genetic variants of 227 glycosylation pathway genes and CM disease-specific survival (CMSS) using genotyping data from two previously published genome-wide association studies. The discovery dataset included 858 CM patients with 95 deaths from The University of Texas MD Anderson Cancer Center, and the replication dataset included 409 CM patients with 48 deaths from Harvard University nurse/physician cohorts. In the multivariable Cox regression analysis, we found that two novel single-nucleotide polymorphisms (SNPs) (ALG6 rs10889417 G>A and GALNTL4 rs12270446 G>C) predicted CMSS, with an adjusted hazards ratios of 0.60 (95% confidence interval = 0.44-0.83 and p = 0.002) and 0.66 (0.52-0.84 and 0.004), respectively. Subsequent expression quantitative trait loci (eQTL) analysis revealed that ALG6 rs10889417 was associated with mRNA expression levels in the cultured skin fibroblasts and whole blood cells and that GALNTL4 rs12270446 was associated with mRNA expression levels in the skin tissues (all p < 0.05). Our findings suggest that, once validated by other large patient cohorts, these two novel SNPs in the glycosylation pathway genes may be useful prognostic biomarkers for CMSS, likely through modulating their gene expression.Item Open Access SINGLE-CHANNEL REAL-TIME DROWSINESS DETECTION BASED ON ELECTROENCEPHALOGRAPHY(2018) Albalawi, Hassan FahadDrowsiness is considered as a major risk factor in workplace injuries and fatalities as much as alcohol. Drowsiness-related accidents tend to be catastrophic. The need of a reliable drowsiness detection system is arising today, as drowsiness is considered as a major cause for many accidents in different sectors. In this thesis, we propose a real-time drowsiness detection system based on a single-channel electroencephalography (EEG). Towards that goal, we introduced three main contribution proposed in this thesis: (1) a real-time drowsiness detection algorithm based on EEG suitable for portable applications with low computational complexity; (2) several novel algorithms to train classifiers that can be implemented on chip with low-power fixed-point arithmetic with extremely small word length; (3) an instantaneous drowsiness detection system suitable for short-time windows of single-channel EEG signal. The proposed real-time drowsiness detection algorithm adopts a cumulative counter to extract important features from 8 different frequency bands. Our experimental results demonstrate that the proposed algorithm is capable of detecting drowsiness with superior accuracy (83.36%) over the conventional method (70.62%). The proposed fixed-point algorithms incorporate the non-idealities (i.e., rounding and overflow) associated with fixed-point arithmetic into the offline training process so that the resulting classifiers are robust to these non-idealities. Our numerical experiments demonstrate that the proposed methods are able to achieve up to 1.67x reduction in the word length compared to the conventional approaches without surrendering any classification accuracy. The instantaneous drowsiness detection algorithm proposed in this work is based on Convolutional Neural Network (CNN). Our experimental results demonstrate that our CNN-based drowsiness detection system is capable of detecting drowsiness in short-time windows (five seconds) with higher accuracy (84.8%) compared to conventional methods (71.0%) and the counter-based method (77.2%). Finally, we briefly discuss few possible research tasks for the future: (1) wearable derive for industrial workers, (2) fixed-point implementation for CNN, and (3) multimode data fusion.
Item Open Access Speaker Diarization with Deep Learning: Refinement, Online Extension and ASR Integration(2023) Wang, WeiqingAs speech remains an essential mode of human communication, the necessity for advanced technologies in speaker diarization has risen significantly. Speaker diarization is the process of accurately annotating individual speakers within an audio segment, and this dissertation explores within this domain, systematically addressing three prevailing challenges through intertwined strands of investigation.
Initially, we focus on the intricacies of overlapping speech and refine the conventional diarization systems with the sequential information integrated. Our approach not only recognizes these overlapping segments but also discerns the distinct speaker identities contained within, ensuring that each speaker is precisely categorized.
Transitioning from the challenge of overlapping speech, we then address the pressing need for real-time speaker diarization. In response to the growing need for low-latency applications in various fields, such as smart agents and transcription services, our research adapts traditional systems, enhancing them to function seamlessly in real-time applications without sacrificing accuracy or efficiency.
Lastly, we turn our attention to the vast reservoir of the potential that lies within contextual and textual data. Incorporating both audio and text data into speaker diarization not only augments the system's ability to distinguish speakers but also leverages the rich contextual cues often embedded in conversations, further improving the overall diarization performance.
Through a coherent and systematic exploration of these three pivotal areas, the dissertation offers substantial contributions to the field of speaker diarization. The research navigates through the challenges of overlapping speech, real-time application demands, and the integration of contextual data, ultimately presenting a refined, reliable, and efficient speaker diarization system poised for application in diverse and dynamic communication environments.
Item Open Access Speaker Representation Learning under Self-supervised and Knowledge Transfer Setting(2023) Cai, DanweiSpeaker representation learning transforms speech signals into informative vectors, underpinning many audio applications. However, deep neural networks (DNNs), pivotal in this domain, falter with limited labeled data.
To overcome this, the thesis presents two primary strategies: self-supervised learning and knowledge transfer from automatic speech recognition (ASR). We introduce a two-stage self-supervised framework utilizing unlabeled data. The first stage focuses on representation learning, while the second integrates clustering and discriminative training. This framework is further streamlined by introducing the self-supervised reflective learning approach, central to which is self-supervised knowledge distillation, optimized to mitigate label noise effects. This approach significantly improves self-supervised speaker representation quality.
Leveraging the relationship between ASR and speaker verification, transfer learning methods are explored to use limited training data efficiently. Techniques include initializing with ASR-pretrained encoders, ASR-based knowledge distillation, and a speaker adaptor converting ASR features to speaker-specific ones.
Additionally, the thesis investigates voice conversion spoofing countermeasures, aiming to detect attacker identities behind conversions.
In essence, this research offers advancements in speaker representation learning, tackling data constraints, and enhancing security against voice spoofing, ultimately fortifying audio applications.
Item Open Access The fungal mycobiome promotes pancreatic oncogenesis via activation of MBL(Nature, 2019-10-10) Aykut, Berk; Pushalkar, Smruti; Chen, Ruonan; Li, Qianhao; Abengozar, Raquel; Kim, Jacqueline I; Shadaloey, Sorin A; Wu, Dongling; Preiss, Pamela; Verma, Narendra; Guo, Yuqi; Saxena, Anjana; Vardhan, Mridula; Diskin, Brian; Wang, Wei; Leinwand, Joshua; Kurz, Emma; Kochen Rossi, Juan A; Hundeyin, Mautin; Zambrinis, Constantinos; Li, Xin; Saxena, Deepak; Miller, GeorgeItem Open Access The Pancreatic Cancer Microbiome Promotes Oncogenesis by Induction of Innate and Adaptive Immune Suppression(Cancer Discovery, 2018-04-01) Pushalkar, Smruti; Hundeyin, Mautin; Daley, Donnele; Zambirinis, Constantinos P; Kurz, Emma; Mishra, Ankita; Mohan, Navyatha; Aykut, Berk; Usyk, Mykhaylo; Torres, Luisana E; Werba, Gregor; Zhang, Kevin; Guo, Yuqi; Li, Qianhao; Akkad, Neha; Lall, Sarah; Wadowski, Benjamin; Gutierrez, Johana; Kochen Rossi, Juan Andres; Herzog, Jeremy W; Diskin, Brian; Torres-Hernandez, Alejandro; Leinwand, Josh; Wang, Wei; Taunk, Pardeep S; Savadkar, Shivraj; Janal, Malvin; Saxena, Anjana; Li, Xin; Cohen, Deirdre; Sartor, R Balfour; Saxena, Deepak; Miller, GeorgeAbstract We found that the cancerous pancreas harbors a markedly more abundant microbiome compared with normal pancreas in both mice and humans, and select bacteria are differentially increased in the tumorous pancreas compared with gut. Ablation of the microbiome protects against preinvasive and invasive pancreatic ductal adenocarcinoma (PDA), whereas transfer of bacteria from PDA-bearing hosts, but not controls, reverses tumor protection. Bacterial ablation was associated with immunogenic reprogramming of the PDA tumor microenvironment, including a reduction in myeloid-derived suppressor cells and an increase in M1 macrophage differentiation, promoting TH1 differentiation of CD4+ T cells and CD8+ T-cell activation. Bacterial ablation also enabled efficacy for checkpoint-targeted immunotherapy by upregulating PD-1 expression. Mechanistically, the PDA microbiome generated a tolerogenic immune program by differentially activating select Toll-like receptors in monocytic cells. These data suggest that endogenous microbiota promote the crippling immune-suppression characteristic of PDA and that the microbiome has potential as a therapeutic target in the modulation of disease progression. Significance: We found that a distinct and abundant microbiome drives suppressive monocytic cellular differentiation in pancreatic cancer via selective Toll-like receptor ligation leading to T-cell anergy. Targeting the microbiome protects against oncogenesis, reverses intratumoral immune tolerance, and enables efficacy for checkpoint-based immunotherapy. These data have implications for understanding immune suppression in pancreatic cancer and its reversal in the clinic. Cancer Discov; 8(4); 403–16. ©2018 AACR. See related commentary by Riquelme et al., p. 386. This article is highlighted in the In This Issue feature, p. 371