Figure 1. Overview of median-supplement methods

Slides:



Advertisements
Similar presentations
Cancer cachexia by J.N. Gordon, S.R. Green, and P.M. Goggin QJM Volume 98(11): October 17, 2005 © The Author Published by Oxford University.
Advertisements

From: A Regularist Approach to Mechanistic Type-Level Explanation
From: An open day in the metric space
Figure 2: Algorithm for detecting early warning signal of a critical transition. From: Detecting tissue-specific early warning signals for complex diseases.
Figure 1. Heat Map Summarizing Question Types by Agency
Figure 1 The spatial and temporal distribution of diarrhoea rates
From: Global Banking: Recent Developments and Insights from Research*
Figure 2. Quality of life (QOL) scores for functional scale items
Fig. 1 Graphical representation
Figure 1. FinHER dataset: distribution of tumor-infiltrating lymphocytes in breast cancer according to the (A) three breast cancer subtypes and (B) HER2.
Figure 2 Effects of political reforms on tariff reduction and political approval rating. From: TPP negotiations and political economy reforms in Japan’s.
Figure 1. Scree plot for the exploratory factor analysis (EFA) of the Death Depression Scale (DDS). From: Development and psychometric evaluation of a.
Figure 2. CONSORT flow diagram.
Fig. 3 Evolutionary game types as a function of relatedness and synergy where each plot is a different social group size. The payoffs used to generate.
Figure 3. Transmission and reception in the BC phase.
From: Software applications for flux balance analysis
Figure 1. Programme provisions, eligibility, evaluation period and geographical coverage of programme over time From: Impact evaluation of free delivery.
Error bars represent the mean (s.e.).
Fig. 1. Map of sample collection sites.
Figure 1 A decision tree diagram for problems and its German correspondences From: Corresponding lexical domains: A new resource for onomasiological.
Figure 3. Examples of tweets classified as ridicule.
Figure 1. Conceptual framework.
Figure 1. Logos appearing in the choice experiment
Figure 1. Selection of papers.
From: Computational methods for identifying miRNA sponge interactions
Figure 1. Conceptual model of well-being related to involvement in theatre. From: Theatre Involvement and Well-Being, Age Differences, and Lessons From.
Figure 1 Individual Lifted from Group Photograph
Figure 1. Orthodontic set-up and location of LLLT or placebo-laser
From: What Drives Local Food Prices
From: Face Recognition is Shaped by the Use of Sign Language
Figure 2. Illustration of minimum cross-sectional area (a) and measurements for oropharynx and soft palate in sagittal plane (b, c). From: Airway volume.
Fig. 1. Geographic locations of Shenzhen and Dongguan
Fig. 1 Proportion of study participants with ideal Cardiovascular Health Metrics by self-rated health. Ideal category of Cardiovascular Health Metrics.
Figure 2. Kaplan–Meier survival analysis in patients
From: Grammatical rhymes in Polish poetry: A quantitative analysis
Fig 1 Respiratory support escalation strategies for study patients
Fig 1 Perfusion index values at different time intervals in patients with successful and failed blocks. A reference line at PI 3.3 is provided. Horizontal.
Figure 1 Flow chart showing the selection of publications identified in the literature search. From: GnRH antagonist versus long agonist protocols in IVF:
Fig. 1. Timeline of the CPAP in Ghana study.
Figure 1. Overall survival of patients receiving alternative medicine (solid lines) vs conventional cancer treatment (dashed lines). Overall survival of.
Example 14. Schubert, Quartet in G Major, D
Figure 2. A consort diagram showing the flowchart of the trial
Figure 1. Single-Tree Model and BART Fits to Simulated Data.
Fig. 1 Selection of patients
Figure 1. Academic productivity and high academic income: top earners vs. the rest of academics. The average number of ‘peer-reviewed article equivalents’
Fig. 1 Sample COMPRENO tree (automatically generated, no manual corrections) From: Text mining War and Peace: Automatic extraction of character traits.
Figure 1. Orthodontic set-up and location of LLLT or placebo-laser
Figure 1. Examples of e-cigarette discussions in social media
From: Estimating the Location of World Wheat Price Discovery
Figure 1. Publication channels used by scholars at the faculty of Arts, 2006–2013. From: Accountability in context: effects of research evaluation systems.
Figure 1: Time points at which sperm samples were analysed for aneuploidy frequencies in controls and cancer patients From: Sperm aneuploidy frequencies.
Figure 1: Logistic regression showing association of prescribing provider and patient panel characteristics with occurrence of a hospitalization or ED.
Figure 3. Visualisation of ESMO-MCB scores for curative and non-curative setting. A & B and 5 and 4 represent the grades with substantial improvement.
NOTE.—Error bars in all figures represent standard errors of the means. From: So Close I Can Almost Sense It: The Interplay between Sensory Imagery and.
Figure 6. Effect of seed MC on the global DNA methylation of Acer platanoides (A) and Acer pseudoplatanus (B) seedlings. Values labeled with different.
Figure 3. Proposed algorithm of treatment for patients with a diagnosis of PI [prepouch ileitis]. From: Prepouch Ileitis After Ileal Pouch-anal Anastomosis:
Figure 1. Impact ratings for prospective memory lapses for younger (~age 30), middle-age (~age 50), and older (~age 65) adults. From: Daily Memory Lapses.
Figure 6. RRs for clinical cure rates stratified by different diseases
Figure 1. Variance partition for the different phases of bud and cambial phenology in black spruce provenances. From: Synchronisms between bud and cambium.
Figure 1. Example of phase shift angles among three different terns where one of them has been taken as a reference. From: Assessment of ELF magnetic fields.
Figure 1. Percentage of participants in each group (holocaust survivors, prewar immigrants, and postwar immigrants) by the main coded strategies. From:
Figure 1. Patterns of HER2–PET/CT confronted with FDG–PET/CT, Maximum intensity projection. Lesion uptake was considered pertinent when visually higher.
Fig. 5 Detailed classifier results for tweets (a) J48, (b) SVM, and (c) NB From: A sentiment analysis system for social media using machine learning techniques:
Fig. 1 Languages used for SA
Figure 4. Influence of Clerk Ideology on Justice Voting.
Figure 1. Non-biochemical recurrence rate for the entire population (n = 122). From: Salvage radiation therapy for prostate cancer patients after prostatectomy.
Somi Jacob and Christian Bach
Unless provided in the caption above, the following copyright applies to the content of this slide: Published on behalf of the European Society of Cardiology.
Figure 1: Trade shares of South Korea's major trading partners (% of South Korea's total trade in goods) Figure 1: Trade shares of South Korea's major.
Figure 1. Forest plot of lung cancer mortality in LDCT trials.
Presentation transcript:

Figure 1. Overview of median-supplement methods Figure 1. Overview of median-supplement methods. The scalar multiplication of the median expression of each gene and a corresponding column vector of a random matrix are aggregated to the gene expressions to form a median-supplement data. The random matrix is generated using a Latin Hypercube. A model for inferring receptor status of a new patient is constructed from the median-supplement data. From: Machine learning approaches to decipher hormone and HER2 receptor status phenotypes in breast cancer Brief Bioinform. Published online October 16, 2017. doi:10.1093/bib/bbx138 Brief Bioinform | © The Author 2017. Published by Oxford University Press. All rights reserved. For Permissions, please email: journals.permissions@oup.com

Figure 2. Accuracy of machine learning methods to identify HER2 status of breast cancer patients. (A) Performance on HER2 receptor data from a group of 162 instances of breast cancer patients. (B) Performance on HER2 receptor data from a group of 806 instances of breast cancer patients. Bars represent the rate of correctly identifying HER2 status of breast cancer patient. SVM is support vector machine, logistic is logistic regression, BN(RHC) is Bayesian network with repeated hill climbing search, BN(SA) is Bayesian network with simulated annealing, NB is Naive Bayes, RT is Random Trees, RF is Random Forest, MNB is median-supplement Naive Bayes and MRF is median-supplement Random Forest. From: Machine learning approaches to decipher hormone and HER2 receptor status phenotypes in breast cancer Brief Bioinform. Published online October 16, 2017. doi:10.1093/bib/bbx138 Brief Bioinform | © The Author 2017. Published by Oxford University Press. All rights reserved. For Permissions, please email: journals.permissions@oup.com

Figure 3. Accuracy of machine learning methods to identify PR status phenotype of breast cancer patients. (A) Performance on PR data from a group of 162 instances of breast cancer patients. (B) Performance on PR data from a group of 1146 instances of breast cancer patients. Bars represent the rate of correctly identifying PR status phenotype of breast cancer patient. SVM is support vector machine, logistic is logistic regression, BN(RHC) is Bayesian network with repeated hill climbing search, BN(SA) is Bayesian network with simulated annealing (excluded in (B) because of the size of data), NB is Naive Bayes, RT is Random Trees, RF is Random Forest, MNB is median-supplement Naive Bayes and MRF is median-supplement Random Forest. From: Machine learning approaches to decipher hormone and HER2 receptor status phenotypes in breast cancer Brief Bioinform. Published online October 16, 2017. doi:10.1093/bib/bbx138 Brief Bioinform | © The Author 2017. Published by Oxford University Press. All rights reserved. For Permissions, please email: journals.permissions@oup.com

Figure 4. Accuracy of machine learning methods to identify ER status phenotype of breast cancer patients. (A) Performance on ER data from a group of 162 instances of breast cancer patients. (B) Performance on ER data from a group of 1149 instances of breast cancer patients. Bars represent the rate of correctly identifying ER status of breast cancer patient. SVM is support vector machine, logistic is logistic regression, BN(RHC) is Bayesian network with repeated hill climbing search, BN(SA) is Bayesian network with simulated annealing (excluded in (B) because of the size of data), NB is Naive Bayes, RT is Random Trees, RF is Random Forest, MNB is median-supplement Naive Bayes and MRF is Median-supplement Random Forest. From: Machine learning approaches to decipher hormone and HER2 receptor status phenotypes in breast cancer Brief Bioinform. Published online October 16, 2017. doi:10.1093/bib/bbx138 Brief Bioinform | © The Author 2017. Published by Oxford University Press. All rights reserved. For Permissions, please email: journals.permissions@oup.com

Figure 5. Effects of different samples of receptor status phenotypes on the performances of median-supplement methods. MNB is median-supplement Naive Bayes and MRF is median-supplement Random Forest. From: Machine learning approaches to decipher hormone and HER2 receptor status phenotypes in breast cancer Brief Bioinform. Published online October 16, 2017. doi:10.1093/bib/bbx138 Brief Bioinform | © The Author 2017. Published by Oxford University Press. All rights reserved. For Permissions, please email: journals.permissions@oup.com