Screening Breast MRI and Gadolinium Deposition: Cause for Concern?

Gadolinium-based contrast agents (GBCAs) have been used worldwide for over 30 years and have enabled lifesaving diagnoses. Contrast-enhanced breast MRI is frequently used as supplemental screening for women with an elevated lifetime risk of breast cancer. Data have emerged that indicate a fractional amount of administered gadolinium is retained in the bone, skin, solid organs, and brain tissues of patients with normal renal function, although there are currently no reliable data regarding the clinical or biological significance of this retention. Linear GBCAs are associated with a higher risk of gadolinium retention than macrocyclic agents. Over the course of their lives, screened women may receive high cumulative doses of GBCA. Therefore, as breast MRI screening utilization increases, thoughtful use of GBCA is indicated in this patient population.

CINmetrics: An R package for chromosomal instability analysis

Genomic instability has been an important hallmark in cancer and more recently in neurodegenrative diseases. Chromosomal instability, as a measure of genomic instability, has been used to characterize clinical and biological phenotypes associated with these diseases by measuring structural and numerical chromosomal alterations. There have been multiple Chromosomal Instability Scores developed across many studies in the literature; however, these scores have not been compared because of a lack of single tool available to calculate these various metrics. Here, we provide an R package CINmetrics, that calculates six different chromosomal instability scores and allows direct comparison between them. We also show how these scores differ by applying CINmetrics to breast cancer data from The Cancer Genome Atlas (TCGA). Availability: The package is available on CRAN at https://cran.r-project.org/package=CINmetrics and on github at https://github.com/lasseignelab/CINmetrics

Empowering study of breast cancer data with application of artificial intelligence technology: promises, challenges, and use cases

In healthcare, artificial intelligence (AI) technologies have the potential to create significant value by improving time-sensitive outcomes while lowering error rates for each patient. Diagnostic images, clinical notes, and reports are increasingly generated and stored in electronic medical records. This heterogeneous data presenting us with challenges in data analytics and reusability that is by nature has high complexity, thereby necessitating novel ways to store, manage and process, and reuse big data. This presents an urgent need to develop new, scalable, and expandable AI infrastructure and analytical methods that can enable healthcare providers to access knowledge for individual patients, yielding better decisions and outcomes. In this review article, we briefly discuss the nature of data in breast cancer study and the role of AI for generating “smart data” which offer actionable information that supports the better decision for personalized medicine for individual patients. In our view, the biggest challenge is to create a system that makes data robust and smart for healthcare providers and patients that can lead to more effective clinical decision-making, improved health outcomes, and ultimately, managing the healthcare outcomes and costs. We highlight some of the challenges in using breast cancer data and propose the need for an AI-driven environment to address them. We illustrate our vision with practical use cases and discuss a path for empowering the study of breast cancer databases with the application of AI and future directions.

TP8.1.7 The utility of a timeline and episode structured breast cancer data system to study outcomes following neoadjuvant chemotherapy for cases stratified by HER2 status

Background Breast neoplasia displays complex patterns of whole-of-life disease progression, which are difficult to study using legacy data systems. Our timeline- and episode-structured breast cancer data set of 20,000 records allows direct visualisation of the entire documentary record of every patient. The embedded data mining module permits research into a wide range of patient cohorts by pathology, treatment and outcome. Methods We selected the cohort of patients aged between 15 and 75 with HER-2 –ve and HER-2 +ve breast cancer who were treated with neoadjuvant chemotherapy (NAC), with or without anti-HER2 therapy between 2002 and 2019. We also studied the patterns and time intervals (in months) of disease progression and response to treatment from primary diagnosis, through loco-regional recurrence and distant metastasis to final outcome. Results Of 301 women with confirmed early stage breast cancer were treated with NAC over that time, 186 had HER2- and 115 had HER2+ tumours. The patterns and intervals of disease progression, as displayed on the Master Lifetrack, were mapped and measured for every patient. The proportions of patients with Her2+ve tumours receiving trastuzumab and analogues, and the tumour responses to treatment, were audited. The underlying data set was validated by review of the original records. Conclusions The whole-of-life timeline structured cancer data system introduces a new direction for clinical data visualisation, record management and user utility in surgical practice. This study validates the model as a tool for the better understanding of treatment effects and longitudinal behaviours in any selected range of cancer phenotypes.

SP8.1.4 Design features of the EPR-integrated Southampton Breast Cancer Data System (SBCDS), with timeline structured whole-of-life records on 20,000 sequential cases since the 1970s

Introduction The digitisation of the electronic patient record (EPR) provides transformative opportunities for data visualisation. The synchronised timeline and iconographic interface permits the whole-of-life display, navigation and interpretation of all documents and reports of each and every EPR on a single screen, thus substantially facilitating clinical research. Methods Since 2010, we have conceived, programmed and iterated a radical interface, UHS Lifelines, within our Trust EPR using agile methodology. It is live for >2.5M record sets, and enriched with cellular pathology records back to 1990. We have integrated this interface into a unique, HTML-enabled, dynamic and continually updated database for the recording of treatments and pathologies of all cases of breast neoplasia from our current and historic record sets. Results As of January 2021, our data system contains ∼20,000 sequential whole of life records of patients with breast neoplasia, including ∼15,000 locally diagnosed and ∼ 5,000 externally referred cases. The unique Cancer Lifetrack timelines displays the disease course of every case from primary diagnosis, through loco-regional recurrence, to distant metastasis, other morbid cancers and cause of death, where relevant. An integral data mining system permits a wide range of analyses. Conclusions We believe our Breast Cancer Data System to be the first-in-class exemplar of a new and proven approach to clinical data visualisation. It permits near-instantaneous oversight and real time updating of every patient record in the system. We recognise its potential application for the whole-of-life study of all chronic diseases of childhood and adulthood as the model is more widely adopted.

Hyperparameter Tuning and Pipeline Optimization via Grid Search Method and Tree-Based AutoML in Breast Cancer Prediction

Automated machine learning (AutoML) has been recognized as a powerful tool to build a system that automates the design and optimizes the model selection machine learning (ML) pipelines. In this study, we present a tree-based pipeline optimization tool (TPOT) as a method for determining ML models with significant performance and less complex breast cancer diagnostic pipelines. Some features of pre-processors and ML models are defined as expression trees and optimal gene programming (GP) pipelines, a stochastic search system. Features of radiomics have been presented as a guide for the ML pipeline selection from the breast cancer data set based on TPOT. Breast cancer data were used in a comparative analysis of the TPOT-generated ML pipelines with the selected ML classifiers, optimized by a grid search approach. The principal component analysis (PCA) random forest (RF) classification was proven to be the most reliable pipeline with the lowest complexity. The TPOT model selection technique exceeded the performance of grid search (GS) optimization. The RF classifier showed an outstanding outcome amongst the models in combination with only two pre-processors, with a precision of 0.83. The grid search optimized for support vector machine (SVM) classifiers generated a difference of 12% in comparison, while the other two classifiers, naïve Bayes (NB) and artificial neural network—multilayer perceptron (ANN-MLP), generated a difference of almost 39%. The method’s performance was based on sensitivity, specificity, accuracy, precision, and receiver operating curve (ROC) analysis.

Research on SPDTRS-PNN Based Intelligent Assistant Diagnosis for Breast Cancer

Background:Breast cancer is the second dangerous cancer in the world. How to identify breast cancer quickly and accurately is of great help to the treatment of breast cancer. Breast cancer data often contains more redundant information. Redundant information makes the breast cancer auxiliary diagnosis less accurate and time-consuming. Dimension reduction algorithm combined with machine learning can solve these problems well. Methods:This paper proposes the single-parameter decision-theoretic rough set (SPDTRS) combined with the probability neural network (PNN) model for breast cancer diagnosis. We structure fifteen models by combining five dimensionality reduction algorithms with three classification algorithms. We compared the accuracy and test time of fifteen models under different parameters or dimensions. We find that when the parameter value of SPDTRS is 2.5, the classification effect of SPDTRS combined with PNN is better. At this point, the number of 30 attributes of the original breast cancer data dropped to 12. Then the SPDTRS-PNN model is further optimized. We compared the accuracy and test time of the model under different SPREAD values in PNN, and established a better SPDTRS-PNN model.Result:We find the parameter value of SPDTRS is 2.5 and the SPREAD value is 0.75, the accuracy of the SPDTRS-PNN model training set is 99.25%, the accuracy of the test set is 97.04%, and the test time is 0.093s.Conclusion:The experimental results show that the SPDTRS-PNN model can improve the accuracy of breast cancer recognition, reduce the time required for diagnosis, and achieve rapid and accurate breast cancer diagnosis.

Correlation-based joint feature screening for semi-competing risks outcomes with application to breast cancer data

Ultrahigh-dimensional gene features are often collected in modern cancer studies in which the number of gene features [Formula: see text] is extremely larger than sample size [Formula: see text]. While gene expression patterns have been shown to be related to patients’ survival in microarray-based gene expression studies, one has to deal with the challenges of ultrahigh-dimensional genetic predictors for survival predicting and genetic understanding of the disease in precision medicine. The problem becomes more complicated when two types of survival endpoints, distant metastasis-free survival and overall survival, are of interest in the study and outcome data can be subject to semi-competing risks due to the fact that distant metastasis-free survival is possibly censored by overall survival but not vice versa. Our focus in this paper is to extract important features, which have great impacts on both distant metastasis-free survival and overall survival jointly, from massive gene expression data in the semi-competing risks setting. We propose a model-free screening method based on the ranking of the correlation between gene features and the joint survival function of two endpoints. The method accounts for the relationship between two endpoints in a simply defined utility measure that is easy to understand and calculate. We show its favorable theoretical properties such as the sure screening and ranking consistency, and evaluate its finite sample performance through extensive simulation studies. Finally, an application to classifying breast cancer data clearly demonstrates the utility of the proposed method in practice.