Large Clinical Datasets Research Articles

A requirement for p120-catenin in the metastasis of invasive ductal breast cancer.

We report here the effects of targeted p120-catenin (encoded by CTNND1; hereafter denoted p120) knockout (KO) in a PyMT mouse model of invasive ductal (mammary) cancer (IDC). Mosaic p120 ablation had little effect on primary tumor growth but caused significant pro-metastatic alterations in the tumor microenvironment, ultimately leading to a marked increase in the number and size of pulmonary metastases. Surprisingly, although early effects of p120-ablation included decreased cell-cell adhesion and increased invasiveness, cells lacking p120 were almost entirely unable to colonized distant metastatic sites in vivo The relevance of this observation to human IDC was established by analysis of a large clinical dataset of 1126 IDCs. As reported by others, p120 downregulation in primary IDC predicted worse overall survival. However, as in the mice, distant metastases were almost invariably p120 positive, even in matched cases where the primary tumors were p120 negative. Collectively, our results demonstrate a strong positive role for p120 (and presumably E-cadherin) during metastatic colonization of distant sites. On the other hand, downregulation of p120 in the primary tumor enhanced metastatic dissemination indirectly via pro-metastatic conditioning of the tumor microenvironment.

Kernel density estimates for sepsis classification

ObjectiveSevere sepsis is a leading cause of intensive care unit (ICU) admission, length of stay, mortality, and cost. systemic inflammatory response syndrome (SIRS) and organ failure due to infection define it, but also make it hard to diagnose. Early diagnosis reduces morbidity, mortality and cost, and diagnosis is often significantly delayed due to a lack of effective biomarkers. This research employs kernel density estimation (KDE) methods fusing a personalized, model-based insulin sensitivity (SI) metric with standard bedside measures of: temperature, heart rate, respiratory rate, systolic and diastolic blood pressure, and SIRS, as these measures are available hourly or more frequently. MethodsModel-based SI is a derived metric, identified using clinical data and a clinically validated metabolic model. The KDE classifier discriminates severe sepsis and septic shock from moderate sepsis using accepted consensus sepsis scores. A best case in-sample estimate, a worst case independent cross validation estimate, and an accepted .632 bootstrap estimate are calculated to assess performance using multi-level likelihood ratios, and sensitivity and specificity. Performance is assessed against clinically and statistically defined thresholds denoted for the minimum acceptable level as: “high accuracy, often providing useful information, and clinical significance,” and similar definitions for greater or lesser quality. ResultsThe .632 bootstrap estimate performs near clinically defined levels of high accuracy, often providing useful information, and clinical significance based on sensitivity, specificity, and multilevel likelihood ratios. Conclusion and significanceThe classifier created and this overall approach is useful for clinical decision making in diagnosing severe sepsis and septic shock in real time, for both case and control hours. However, improvements could be made with larger clinical data sets.

Updating the International Index of Erectile Function: Evaluation of a Large Clinical Data Set

IntroductionThe International Index of Erectile Function (IIEF) is the predominant patient-reported outcomes instrument for assessing male sexual function. There are obvious problems with the use of the IIEF in the assessment of an individual patient, such as for men who use injections and men who do not engage in intercourse. AimThe aim of the current study is to redesign the erectile function domain of the IIEF (IIEF6) to more accurately assess the individual patient. MethodsIn an observational study of men undergoing treatment for prostate cancer at a tertiary care institution, including 24,732 questionnaires completed by 6,780 individuals, IIEF6 scores were compared for patients using and not using erectile aids. Men not engaging in sexual intercourse were asked to describe the reason. Main Outcome MeasureThe main outcome we were seeking was the IIEF6 scores. ResultsMean scores before erectile aids items were added was 17.7 compared with predicted scores of 18.3 vs 16.7 if patients reported their function with vs without the use of aids. No intercourse was reported for 35% of surveys. Reasons given were lack of ability or confidence in 53%, lack of willing and available partner in 28%, “other” in 17% (including respondent’s or partner’s health issues, low libido, preference for nonpenetrative sex or for sex with men). Doubling the sum of the 3 nonintercourse IIEF6 questions had excellent properties (difference of 0.06, limits of agreement -3.10 to 3.22). Clinical ImplicationsErectile function instruments must include items about erectile aids. Men who report that they have not attempted intercourse should not be assumed to have erectile dysfunction, but should be asked the reason why. For men who report lack of opportunity or preference for intercourse, the score of the 3 nonintercourse IIEF6 questions should be doubled. Strengths & LimitationsThis is a large study of patients in a real-world setting. Although the study only includes radical prostatectomy patients, and although the study cohort is not fully representative of the US prostate cancer population as a whole, these issues would not affect the key findings. ConclusionThe IIEF6 can be redesigned to better assess the individual patient.Vickers AJ, Tin AL, Singh K, et al. Updating the International Index of Erectile Function: Evaluation of a Large Clinical Data Set. J Sex Med 2020;17:126–132.

LongSil: an Evaluation Metric to Assess Quality of Clustering Longitudinal Clinical Data.

Longitudinal disease subtyping is an important problem within the broader scope of computational phenotyping. In this article, we discuss several data-driven unsupervised disease subtyping methods to obtain disease subtypes from longitudinal clinical data. The methods are analyzed in the context of chronic kidney disease, one of the leading health problems, both in the USA and worldwide. To provide a quantitative comparison of the different methods, we propose a novel evaluation metric that measures the cluster tightness and degree of separation between the various clusters produced by each method. Comparative results for two significantly large clinical datasets are provided, along with key insights that are possible due to the proposed evaluation metric.

Consistency of quantitative electroencephalography features in a large clinical data set

Objective. Despite their increasing use and public health importance, little is known about the consistency and variability of the quantitative features of baseline electroencephalography (EEG) measurements in healthy individuals and populations. This study aims to investigate population consistency of EEG features. Approach. We propose a non-parametric method of evaluating consistency of commonly used EEG features based on counts of non-significant statistical tests using a large data set. We first replicate stationarity results of absolute band powers using coefficients of variation. We then determine feature stationarity, intra-subject consistency, inter-subject consistency, and intra- versus inter-subject consistency across different epoch lengths for 30 features. Main results. We find in general that features with normalizing constants are more stationary. We also find entropy, median, skew, and kurtosis of EEG to behave as baseline EEG metrics. However, other spectral and signal shape features have stronger intra-subject consistency and thus are better for distinguishing individuals. Significance. These results provide data-driven non-parametric methods of identifying EEG features and their spatial characteristics ideal for various EEG applications, and determining future EEG feature consistencies using an existing EEG data set.

1574. Predictive Ability and Bias of Vancomycin Population PK Models in an Obese Adult Population

BackgroundAccurate dosing of vancomycin is difficult due to high inter-individual variability of vancomycin pharmacokinetics (PK), and is particularly challenging in obese patients. Vancomycin is hydrophilic, yet total body weight (TBW) has traditionally been used for dosing in the general population, and also into the obese population. The aim of this study was to evaluate the performance of published vancomycin PK models in a large set of routine clinical data obtained from an obese population.MethodsDe-identified data were available from 1717 courses of vancomycin administered to obese adults (BMI ≥ 30) from hospitals across the United States, EU, and Australia. Three population PK models, Buelga et al. (2005), Goti et al. (2018), and an obese-specific model, Adane et al. (2015), were used to predict plasma concentrations at the time of each recorded vancomycin assay, and their accuracy and bias were compared. Goodness of fit at both the population and individual level was assessed, and elastic net regression was used to identify any sources of predictive error in the obese-specific model. Model parameters for each model were then re-estimated, and a variety of body size metrics were evaluated.ResultsThe Buelga et al. one-compartment model had the best predictive ability (Table 1). In all models, bias (calculated as MME; mean per-patient mean predictive error) by obesity class was observed at both the population and individual levels, and unexpectedly was largest in the obese-specific model. In the obese-specific model, predictive error correlated with the use of TBW as a model covariate. A set of models derived from Adane et al. model were then developed to correct for weight. Using ideal body weight (IBW) on Vd and no correction for weight on CL provided the best fit. The derived model accounted for 81% of variance in plasma concentration and exhibited negligible bias by obesity class (population MME = -0.75 (i), -0.06 (ii), and -0.50 (iii) mg/L; individual MME = −0.20 (i), −0.02 (ii), and −0.02 (iii) mg/L).ConclusionExisting vancomycin population PK models for use in the obese population are biased in higher obesity classes due to the use of total body weight. A novel population PK model developed using ideal body weight exhibits negligible bias across obesity classes as well as improved predictive ability. Disclosures All authors: No reported disclosures.

Inflammation and depressive phenotypes: evidence from medical records from over 12 000 patients and brain morphology.

Preclinical and human studies suggest an association between chronic inflammation and the development of depressive behaviors. This is proposed to occur through downstream effects of inflammatory cytokines on neuroplasticity, neurogenesis and neurotransmitter function, although the neural correlates remain poorly understood in humans. In Study 1, structural magnetic resonance imaging and serum inflammatory cytokine data were analyzed from 53 psychiatrically healthy female participants. Correlational analyses were conducted between interleukin-6 (IL-6) and volume in a priori regions implicated in the pathophysiology of major depressive disorder (MDD). In Study 2, medical data [including serum inflammatory acute phase reactants (C-reactive protein)] were analyzed for 12 589 participants. Participants were classified as having (n = 2541) v. not having (n = 10 048) probable lifetime MDD using phenotypes derived using machine-learning approaches. Non-parametric analyses compared inflammation between groups, whereas regression analyses probed whether inflammation predicted probable MDD classification while accounting for other variables. In Study 1, significant negative correlations emerged between IL-6 and hippocampal, caudate, putamen and amygdalar volume. In Study 2, the MDD group showed a higher probability of elevated inflammation than the non-MDD group. Moreover, elevated inflammation was a significant predictor of probable MDD classification. Findings indicate that inflammation is cross-sectionally related to reduced volume in brain regions implicated in MDD phenotypes among a sample of psychiatrically healthy women, and is associated with the presence of probable MDD in a large clinical dataset. Future investigations may identify specific inflammatory markers predicting first MDD onset.

STrategically Acquired Gradient Echo (STAGE) imaging, part III: Technical advances and clinical applications of a rapid multi-contrast multi-parametric brain imaging method

One major thrust in radiology today is image standardization with a focus on rapidly acquired quantitative multi-contrast information. This is critical for multi-center trials, for the collection of big data and for the use of artificial intelligence in evaluating the data. Strategically acquired gradient echo (STAGE) imaging is one such method that can provide 8 qualitative and 7 quantitative pieces of information in 5 min or less at 3 T. STAGE provides qualitative images in the form of proton density weighted images, T1 weighted images, T2* weighted images and simulated double inversion recovery (DIR) images. STAGE also provides quantitative data in the form of proton spin density, T1, T2* and susceptibility maps as well as segmentation of white matter, gray matter and cerebrospinal fluid. STAGE uses vendors' product gradient echo sequences. It can be applied from 0.35 T to 7 T across all manufacturers producing similar results in contrast and quantification of the data. In this paper, we discuss the strengths and weaknesses of STAGE, demonstrate its contrast-to-noise (CNR) behavior relative to a large clinical data set and introduce a few new image contrasts derived from STAGE, including DIR images and a new concept referred to as true susceptibility weighted imaging (tSWI) linked to fluid attenuated inversion recovery (FLAIR) or tSWI-FLAIR for the evaluation of multiple sclerosis lesions. The robustness of STAGE T1 mapping was tested using the NIST/NIH phantom, while the reproducibility was tested by scanning a given individual ten times in one session and the same subject scanned once a week over a 12-week period. Assessment of the CNR for the enhanced T1W image (T1WE) showed a significantly better contrast between gray matter and white matter than conventional T1W images in both patients with Parkinson's disease and healthy controls. We also present some clinical cases using STAGE imaging in patients with stroke, metastasis, multiple sclerosis and a fetus with ventriculomegaly. Overall, STAGE is a comprehensive protocol that provides the clinician with numerous qualitative and quantitative images.

Usability Testing of an Interactive Dashboard for Surgical Quality Improvement in a Large Congenital Heart Center.

Interactive data visualization and dashboards can be an effective way to explore meaningful patterns in large clinical data sets and to inform quality improvement initiatives. However, these interactive dashboards may have usability issues that undermine their effectiveness. These usability issues can be attributed to mismatched mental models between the designers and the users. Unfortunately, very few evaluation studies in visual analytics have specifically examined such mismatches between these two groups. We aimed to evaluate the usability of an interactive surgical dashboard and to seek opportunities for improvement. We also aimed to provide empirical evidence to demonstrate the mismatched mental models between the designers and the users of the dashboard. An interactive dashboard was developed in a large congenital heart center. This dashboard provides real-time, interactive access to clinical outcomes data for the surgical program. A mixed-method, two-phase study was conducted to collect user feedback. A group of designers (N = 3) and a purposeful sample of users (N = 12) were recruited. The qualitative data were analyzed thematically. The dashboards were compared using the System Usability Scale (SUS) and qualitative data. The participating users gave an average SUS score of 82.9 on the new dashboard and 63.5 on the existing dashboard (p = 0.006). The participants achieved high task accuracy when using the new dashboard. The qualitative analysis revealed three opportunities for improvement. The data analysis and triangulation provided empirical evidence to the mismatched mental models. We conducted a mixed-method usability study on an interactive surgical dashboard and identified areas of improvements. Our study design can be an effective and efficient way to evaluate visual analytics systems in health care. We encourage researchers and practitioners to conduct user-centered evaluation and implement education plans to mitigate potential usability challenges and increase user satisfaction and adoption.

Automatic Prediction of Meningioma Grade Image Based on Data Amplification and Improved Convolutional Neural Network.

Meningioma is the second most commonly encountered tumor type in the brain. There are three grades of meningioma by the standards of the World Health Organization. Preoperative grade prediction of meningioma is extraordinarily important for clinical treatment planning and prognosis evaluation. In this paper, we present a new deep learning model for assisting automatic prediction of meningioma grades to reduce the recurrence of meningioma. Our model is based on an improved LeNet-5 model of convolutional neural network (CNN) and does not require the extraction of the diseased tissue, which can greatly enhance the efficiency. To address the issue of insufficient and unbalanced clinical data of meningioma images, we use an oversampling technique which allows us to considerably improve the accuracy of classification. Experiments on large clinical datasets show that our model can achieve quite high accuracy (i.e., as high as 83.33%) for the classification of meningioma images.

Analysis of a Unique Postmarket Surveillance Dataset That a Glucose Test-Strip Demonstrates no Evidence of Interference and Robust Clinical Accuracy Irrespective of the Prescription Medication Status of a Large Cohort of Patients With Diabetes

Despite a marked increase in polypharmacy in patients with diabetes there have been no thorough evaluations of the impact of polypharmacy on the accuracy of any current blood glucose monitoring (BGM) system. This study evaluated the accuracy of a BGM test-strip with respect to polypharmacy using a large clinical registry dataset. Medication profiles were analyzed for 830 subjects (334 with type 1 [T1D] and 496 with type 2 diabetes [T2D]) attending three hospitals. Blood samples were analyzed to determine clinical accuracy of the BGM test-strip compared to a laboratory comparator. Across the 830 subjects, 473 different medications (41 diabetes and 432 nondiabetes) were recorded. Patients took on average 6.5 (n = 1-23) individual medications and 4 (n = 1-11) unique classes of medication. Clinical accuracy to EN ISO 15197:2015 criteria was met irrespective of increasing average number of individual medication, categorized from 1 to 4, 5 to 8, 9 to 12, and >12 taken per subject (97.7%, 98.4%, 98.1%, and 98.5%, respectively). Clinical accuracy to EN ISO 15197:2015 criteria was also met across 15 classes of medication using the combined dataset (98.1%; 13 003/13 253). Surveillance error grid analysis showed 98.8% (13 079/13 232) of readings presented no clinical risk. No individual class or combination of medication classes impacted clinical accuracy of the BGM test-strip. This comprehensive analysis for this specific test-strip platform demonstrated no evidence of interference and robust clinical accuracy of this test strip, irrespective of the prescription medication status of patients with diabetes.

Psychiatric comorbidity in Intermittent Explosive Disorder

This study examined aspects of psychiatry comorbidity in Intermittent Explosive Disorder (IED) in order to explore the validity of IED in the context of other psychiatric disorders. Data from the National Comorbidity Study – Revised (NCS-R: n = 9,282 adults) and its Adolescent Supplement (NCS-AS: n = 10,148 adolescents) and a large clinical research data set (n = 1640) were analyzed in this study. Mean number of comorbid disorders among current IED participants was similar to that among other Non-IED disorders and comorbidity of IED with Non-IED disorders was similar to comorbidity among Non-IED disorders. When examined together, current IED was significantly comorbid with current bipolar, depressive, anxiety, substance use, and post-traumatic stress disorder, and age of onset of IED generally preceded that of the comorbid disorders. Finally, mean aggression scores were generally highest among those with IED and the comorbid disorder while scores among those with the comorbid disorder alone were generally less than that with IED or with IED and the comorbid disorder. Comorbidity in IED is similar to that in other, more established, Non-IED disorders. The observation that the development of IED precedes the onset of most comorbid disorders and that aggression scores in those with IED and a comorbid disorder are often higher than those with IED alone supports the rationale that a diagnosis of both IED and the comorbid disorder should be made when both are present.

Perspectives from the Innovation and Quality Consortium Induction Working Group on Factors Impacting Clinical Drug-Drug Interactions Resulting from Induction: Focus on Cytochrome 3A Substrates.

A recent publication from the Innovation and Quality Consortium Induction Working Group collated a large clinical data set with the goal of evaluating the accuracy of drug-drug interaction (DDI) prediction from in vitro data. Somewhat surprisingly, comparison across studies of the mean- or median-reported area under the curve ratio showed appreciable variability in the magnitude of outcome. This commentary explores the possible drivers of this range of outcomes observed in clinical induction studies. While recommendations on clinical study design are not being proposed, some key observations were informative during the aggregate analysis of clinical data. Although DDI data are often presented using median data, individual data would enable evaluation of how differences in study design, baseline expression, and the number of subjects contribute. Since variability in perpetrator pharmacokinetics (PK) could impact the overall DDI interpretation, should this be routinely captured? Maximal induction was typically observed after 5-7 days of dosing. Thus, when the half-life of the inducer is less than 30 hours, are there benefits to a more standardized study design? A large proportion of CYP3A4 inducers were also CYP3A4 inhibitors and/or inactivators based on in vitro data. In these cases, using CYP3A selective substrates has limitations. More intensive monitoring of changes in area under the curve over time is warranted. With selective CYP3A substrates, the net effect was often inhibition, whereas less selective substrates could discern induction through mechanisms not susceptible to inhibition. The latter included oral contraceptives, which raise concerns of reduced efficacy following induction. Alternative approaches for modeling induction, such as applying biomarkers and physiologically based pharmacokinetic modeling (PBPK), are also considered. SIGNIFICANCE STATEMENT: The goal of this commentary is to stimulate discussion on whether there are opportunities to optimize clinical drug-drug interaction study design. The overall aim is to reduce, understand and contextualize the variability observed in the magnitude of induction across reported clinical studies. A large clinical CYP3A induction dataset was collected and further analyzed to identify trends and gaps. Reporting individual victim PK data, characterizing perpetrator PK and including additional PK assessments for mixed-mechanism perpetrators may provide insights into how these factors impact differences observed in clinical outcomes. The potential utility of biomarkers and PBPK modeling are discussed in considering future directions.

Quantifying the Association Between Psychotherapy Content and Clinical Outcomes Using Deep Learning

Compared with the treatment of physical conditions, the quality of care of mental health disorders remains poor and the rate of improvement in treatment is slow, a primary reason being the lack of objective and systematic methods for measuring the delivery of psychotherapy. To use a deep learning model applied to a large-scale clinical data set of cognitive behavioral therapy (CBT) session transcripts to generate a quantifiable measure of treatment delivered and to determine the association between the quantity of each aspect of therapy delivered and clinical outcomes. All data were obtained from patients receiving internet-enabled CBT for the treatment of a mental health disorder between June 2012 and March 2018 in England. Cognitive behavioral therapy was delivered in a secure online therapy room via instant synchronous messaging. The initial sample comprised a total of 17 572 patients (90 934 therapy session transcripts). Patients self-referred or were referred by a primary health care worker directly to the service. All patients received National Institute for Heath and Care Excellence-approved disorder-specific CBT treatment protocols delivered by a qualified CBT therapist. Clinical outcomes were measured in terms of reliable improvement in patient symptoms and treatment engagement. Reliable improvement was calculated based on 2 severity measures: Patient Health Questionnaire (PHQ-9) and Generalized Anxiety Disorder 7-item scale (GAD-7), corresponding to depressive and anxiety symptoms respectively, completed by the patient at initial assessment and before every therapy session (see eMethods in the Supplement for details). Treatment sessions from a total of 14 899 patients (10 882 women) aged between 18 and 94 years (median age, 34.8 years) were included in the final analysis. We trained a deep learning model to automatically categorize therapist utterances into 1 or more of 24 feature categories. The trained model was applied to our data set to obtain quantifiable measures of each feature of treatment delivered. A logistic regression revealed that increased quantities of a number of session features, including change methods (cognitive and behavioral techniques used in CBT), were associated with greater odds of reliable improvement in patient symptoms (odds ratio, 1.11; 95% CI, 1.06-1.17) and patient engagement (odds ratio, 1.20, 95% CI, 1.12-1.27). The quantity of nontherapy-related content was associated with reduced odds of symptom improvement (odds ratio, 0.89; 95% CI, 0.85-0.92) and patient engagement (odds ratio, 0.88, 95% CI, 0.84-0.92). This work demonstrates an association between clinical outcomes in psychotherapy and the content of therapist utterances. These findings support the principle that CBT change methods help produce improvements in patients' presenting symptoms. The application of deep learning to large clinical data sets can provide valuable insights into psychotherapy, informing the development of new treatments and helping standardize clinical practice.

Machine learning in orthodontics: Introducing a 3D auto-segmentation and auto-landmark finder of CBCT images to assess maxillary constriction in unilateral impacted canine patients.

To (1) introduce a novel machine learning method and (2) assess maxillary structure variation in unilateral canine impaction for advancing clinically viable information. A machine learning algorithm utilizing Learning-based multi-source IntegratioN frameworK for Segmentation (LINKS) was used with cone-beam computed tomography (CBCT) images to quantify volumetric skeletal maxilla discrepancies of 30 study group (SG) patients with unilaterally impacted maxillary canines and 30 healthy control group (CG) subjects. Fully automatic segmentation was implemented for maxilla isolation, and maxillary volumetric and linear measurements were performed. Analysis of variance was used for statistical evaluation. Maxillary structure was successfully auto-segmented, with an average dice ratio of 0.80 for three-dimensional image segmentations and a minimal mean difference of two voxels on the midsagittal plane for digitized landmarks between the manually identified and the machine learning-based (LINKS) methods. No significant difference in bone volume was found between impaction ([2.37 ± 0.34] [Formula: see text] 104 mm3) and nonimpaction ([2.36 ± 0.35] [Formula: see text] 104 mm3) sides of SG. The SG maxillae had significantly smaller volumes, widths, heights, and depths (P < .05) than CG. The data suggest that palatal expansion could be beneficial for those with unilateral canine impaction, as underdevelopment of the maxilla often accompanies that condition in the early teen years. Fast and efficient CBCT image segmentation will allow large clinical data sets to be analyzed effectively.

Notch pathway in ependymoma RELA-fused subgroup: upregulation and association with cancer stem cells markers expression.

RELA-fused supratentorial (ST) ependymoma (EPN) is an aggressive subgroup with poor prognosis. Considering the putative role of Notch signaling in the maintenance of the cancer stem cells (CSC) phenotype in RELA-fused EPN, we investigated the expression of Notch pathway and its target genes in this subgroup. We also evaluated the effects of two Notch inhibitors (DAPT and RO4929097) on cell proliferation, apoptosis, colony formation, and CSCs markers gene expression on EPN cell line of the RELA-fused subgroup (BXD-1425). In addition, in silico signatures of the Notch genes and CSCs markers were analyzed on a large clinical dataset from GSE64415 study. We found that among the ST-EPN subgroups the Notch signaling (NOTCH1, JAG1, JAG2, and HES4) is specifically activated in the ST-EPN-RELA. Furthermore, treatment of the RELA-fused EPN cell line with the Notch inhibitors impaired the Notch signaling expression and revealed that Notch axis is not essential for cell proliferation and survival in this setting. NOTCH1 expression in ST-EPN was correlated with the CSCs markers VEGFA and L1CAM overexpression and JAG1 expression was correlated with the CCND1 and CDK6 overexpression. In addition, in vitro treatment with Notch inhibitors induced downregulation of CSCs markers. These findings indicate that Notch signaling can be involved in the ST-EPN-RELA CSCs maintenance by modulating the expression of genes responsible for cell phenotype and cell fate.

Robust Maximum Likelihood Estimation

In many applications, statistical estimators serve to derive conclusions from data, for example, in finance, medical decision making, and clinical trials. However, the conclusions are typically dependent on uncertainties in the data. We use robust optimization principles to provide robust maximum likelihood estimators that are protected against data errors. Both types of input data errors are considered: (a) the adversarial type, modeled using the notion of uncertainty sets, and (b) the probabilistic type, modeled by distributions. We provide efficient local and global search algorithms to compute the robust estimators and discuss them in detail for the case of multivariate normally distributed data. The estimator performance is demonstrated on two applications. First, using computer simulations, we demonstrate that the proposed estimators are robust against both types of data uncertainty and provide more accurate estimates compared with classical estimators, which degrade significantly, when errors are encountered. We establish a range of uncertainty sizes for which robust estimators are superior. Second, we analyze deviations in cancer radiation therapy planning. Uncertainties among plans are caused by patients’ individual anatomies and the trial-and-error nature of the process. When analyzing a large set of past clinical treatment data, robust estimators lead to more reliable decisions when applied to a large set of past treatment plans.

Electronic Health Informatics Data To Describe Clearance Dynamics of Hepatitis B Surface Antigen (HBsAg) and e Antigen (HBeAg) in Chronic Hepatitis B Virus Infection.

HBsAg and HBeAg have gained traction as biomarkers of control and clearance during chronic hepatitis B virus infection (CHB). Improved understanding of the clearance correlates of these proteins could help inform improvements in patient-stratified care and advance insights into the underlying mechanisms of disease control, thus underpinning new cure strategies. We collected electronic clinical data via an electronic pipeline supported by the National Institute for Health Research Health Informatics Collaborative (NIHR HIC), adopting an unbiased approach to the generation of a robust longitudinal data set for adults testing HBsAg positive from a large UK teaching hospital over a 6-year period (2011 to 2016 inclusive). Of 553 individuals with CHB, longitudinal data were available for 319, representing >107,000 weeks of clinical follow-up. Among these 319 individuals, 13 (4%) cleared HBsAg completely. Among these 13, the HBsAg clearance rate in individuals on nucleos(t)ide analogue (NA) therapy (n = 4 [31%]; median clearance time,150 weeks) was similar to that in individuals not on NA therapy (n = 9 [69%]; median clearance time, 157 weeks). Those who cleared HBsAg were significantly older and less likely to be on NA therapy than nonclearers (P = 0.003 and P = 0.001, respectively). Chinese ethnicity was associated with HBeAg positivity (P = 0.025). HBeAg clearance occurred in individuals both on NA therapy (n = 24; median time, 49 weeks) and off NA therapy (n = 19; median time, 52 weeks). Improved insights into the dynamics of these biomarkers can underpin better prognostication and patient-stratified care. Our systematized approach to data collection paves the way for scaling up efforts to harness clinical data to address research questions and support improvements in clinical care.IMPORTANCE Advances in the diagnosis, monitoring, and treatment of hepatitis B virus (HBV) infection are urgently required if we are to meet international targets for elimination by the year 2030. Here we demonstrate how routine clinical data can be harnessed through an unbiased electronic pipeline, showcasing the significant potential for amassing large clinical data sets that can help to inform advances in patient care and provide insights that may help to inform new cure strategies. Our cohort from a large UK hospital includes adults from diverse ethnic groups that have previously been underrepresented in the literature. By tracking two protein biomarkers that are used to monitor chronic HBV infection, we provide new insights into the timelines of HBV clearance, both on and off treatment. These results contribute to improvements in individualized clinical care and may provide important clues into the immune events that underpin disease control.

Machine learning with the TCGA-HNSC dataset: improving usability by addressing inconsistency, sparsity, and high-dimensionality

BackgroundIn the era of precision oncology and publicly available datasets, the amount of information available for each patient case has dramatically increased. From clinical variables and PET-CT radiomics measures to DNA-variant and RNA expression profiles, such a wide variety of data presents a multitude of challenges. Large clinical datasets are subject to sparsely and/or inconsistently populated fields. Corresponding sequencing profiles can suffer from the problem of high-dimensionality, where making useful inferences can be difficult without correspondingly large numbers of instances. In this paper we report a novel deployment of machine learning techniques to handle data sparsity and high dimensionality, while evaluating potential biomarkers in the form of unsupervised transformations of RNA data. We apply preprocessing, MICE imputation, and sparse principal component analysis (SPCA) to improve the usability of more than 500 patient cases from the TCGA-HNSC dataset for enhancing future oncological decision support for Head and Neck Squamous Cell Carcinoma (HNSCC).ResultsImputation was shown to improve prognostic ability of sparse clinical treatment variables. SPCA transformation of RNA expression variables reduced runtime for RNA-based models, though changes to classifier performance were not significant. Gene ontology enrichment analysis of gene sets associated with individual sparse principal components (SPCs) are also reported, showing that both high- and low-importance SPCs were associated with cell death pathways, though the high-importance gene sets were found to be associated with a wider variety of cancer-related biological processes.ConclusionsMICE imputation allowed us to impute missing values for clinically informative features, improving their overall importance for predicting two-year recurrence-free survival by incorporating variance from other clinical variables. Dimensionality reduction of RNA expression profiles via SPCA reduced both computation cost and model training/evaluation time without affecting classifier performance, allowing researchers to obtain experimental results much more quickly. SPCA simultaneously provided a convenient avenue for consideration of biological context via gene ontology enrichment analysis.

Saving bones without risking brain-bisphosphonates and risk of stroke: matched case-control study.

There is conflicting evidence on the link between bisphosphonates and stroke with studies variously showing increased, decreased or unchanged risk. We investigated the association between bisphosphonate treatment and the risk of stroke using a large routine clinical dataset. We used a matched nested case-control study design analysing routinely collected electronic data from patients registered at primary care practices in England participating in the Royal College of General Practitioners Research and Surveillance Centre. Cases were patients aged 18years or over, either living or dead, recorded as having had a stroke in the period 1 January 2005 to 31 March 2016. Each case was matched to one control according to age, sex, general practice attended and calendar time. Data were analysed using Stata, version 14.2. and RStudio, version 1.1.463. Conditional logistic regression was used to determine odds ratios for stroke according to bisphosphonate treatment and duration in cases compared with controls. We adjusted for disease risk groups, cardiovascular risk factors, treatments, smoking status, alcohol consumption, ethnicity, bisphosphonate types, fracture and socioeconomic status using IMD (Index of Multiple Deprivation). We included 31,414 cases of stroke with an equal number of matched controls. Overall, 83.2% of cases and controls were aged 65years or older, and there were similar proportions of females (51.5%) and males (48.5%). Bisphosphonate treatment was not associated with stroke after adjusting for the wide range of confounders considered (OR 0.86, 95% CI 0.62-1.19). We found no association between bisphosphonate treatment and risk of stroke, after adjusting for other confounders.