Dataset Information

Weakly supervised temporal model for prediction of breast cancer distant recurrence.

ABSTRACT: Efficient prediction of cancer recurrence in advance may help to recruit high risk breast cancer patients for clinical trial on-time and can guide a proper treatment plan. Several machine learning approaches have been developed for recurrence prediction in previous studies, but most of them use only structured electronic health records and only a small training dataset, with limited success in clinical application. While free-text clinic notes may offer the greatest nuance and detail about a patient's clinical status, they are largely excluded in previous predictive models due to the increase in processing complexity and need for a complex modeling framework. In this study, we developed a weak-supervision framework for breast cancer recurrence prediction in which we trained a deep learning model on a large sample of free-text clinic notes by utilizing a combination of manually curated labels and NLP-generated non-perfect recurrence labels. The model was trained jointly on manually curated data from 670 patients and NLP-curated data of 8062 patients. It was validated on manually annotated data from 224 patients with recurrence and achieved 0.94 AUROC. This weak supervision approach allowed us to learn from a larger dataset using imperfect labels and ultimately provided greater accuracy compared to a smaller hand-curated dataset, with less manual effort invested in curation.

SUBMITTER: Sanyal J

PROVIDER: S-EPMC8096809 | biostudies-literature | 2021 May

REPOSITORIES: biostudies-literature

ACCESS DATA

Publications

Weakly supervised temporal model for prediction of breast cancer distant recurrence.

Sanyal Josh J Tariq Amara A Kurian Allison W AW Rubin Daniel D Banerjee Imon I

Scientific reports 20210504 1

Efficient prediction of cancer recurrence in advance may help to recruit high risk breast cancer patients for clinical trial on-time and can guide a proper treatment plan. Several machine learning approaches have been developed for recurrence prediction in previous studies, but most of them use only structured electronic health records and only a small training dataset, with limited success in clinical application. While free-text clinic notes may offer the greatest nuance and detail about a pat ...[more]

PMID: 33947927

Similar Datasets

Project description:ObjectiveBreast cancer stands as the most prevalent form of cancer among women globally. This heterogeneous disease exhibits varying clinical behaviors. The stratification of breast cancer patients into risk groups, determined by their metastasis and survival outcomes, is pivotal for tailoring personalized treatments and therapeutic interventions. The pathological sections of radical specimens encompass a diverse range of histological information pertinent to the metastasis and survival of patients. In this study, our objective is to develop a deep learning model utilizing pathological images to predict the metastasis and survival outcomes for breast cancer patients.MethodsThis study utilized pathological sections from 204 radical mastectomy specimens obtained between January 2013 and December 2014 at the Second Affiliated Hospital of the Medical College of Zhejiang University. The 204 pathological slices were scanned and transformed into whole slide imaging (WSI), with manual labeling of all tumor areas. The WSI was then partitioned into smaller tiles measuring 512 × 512 pixels. Three networks, namely Densely Connected Convolutional Network 121 (DenseNet121), Residual Network (ResNet50), and Inception_v3, were assessed. Subsequently, we combined patch-level predictions, probability histograms, and Term Frequency-Inverse Document Frequency (TF-IDF) features to create comprehensive participants representations. These features served as the foundational input for developing a machine learning algorithm for metastasis analysis and a Cox regression model for survival analysis.ResultOur results show that the Inception_v3 model shows a particularly robust patch recognition ability for estrogen receptor (ER) recognition. Our pathological model shows high accuracy in predicting tumor regions. The train area under the curve (AUC) of the Inception_v3 model based on supervised learning is 0.975, which is higher than the model established by weakly supervised learning. But the AUC of the metastasis prediction in training and testing sets is higher than value based on supervised learning. Furthermore, the C-index of the survival prediction model is 0.710 in the testing sets, which is also better than the value by supervised learning.ConclusionOur study demonstrates the significant potential of deep learning models in predicting breast cancer metastasis and prognosis, with the pathomic model showing high accuracy in identifying tumor areas and ER status. The integration of clinical features and pathomics signature into a nomogram further provides a valuable tool for clinicians to make individualized treatment decisions.

Project description:CanAssist-Breast (CAB) is an immunohistochemistry (IHC)-based prognostic test for early-stage Hormone Receptor (HR+)-positive breast cancer patients. CAB uses a Support Vector Machine (SVM) trained algorithm which utilizes expression levels of five biomarkers (CD44, ABCC4, ABCC11, N-Cadherin, and Pan-Cadherin) and three clinical parameters such as tumor size, grade, and node status as inputs to generate a risk score and categorizes patients as low- or high-risk for distant recurrence within 5 years of diagnosis. In this study, we present clinical validation of CAB. CAB was validated using a retrospective cohort of 857 patients. All patients were treated either with endocrine therapy or chemoendocrine therapy. Risk categorization by CAB was analyzed by calculating Distant Metastasis-Free Survival (DMFS) and recurrence rates using Kaplan-Meier survival curves. Multivariate analysis was performed to calculate Hazard ratios (HR) for CAB high-risk vs low-risk patients. The results showed that Distant Metastasis-Free Survival (DMFS) was significantly different (P-0.002) between low- (DMFS: 95%) and high-risk (DMFS: 80%) categories in the endocrine therapy treated alone subgroup (n = 195) as well as in the total cohort (n = 857, low-risk DMFS: 95%, high-risk DMFS: 84%, P < 0.0001). In addition, the segregation of the risk categories was significant (P = 0.0005) in node-positive patients, with a difference in DMFS of 12%. In multivariate analysis, CAB risk score was the most significant predictor of distant recurrence with hazard ratio of 3.2048 (P < 0.0001). CAB stratified patients into discrete risk categories with high statistical significance compared to Ki-67 and IHC4 score-based stratification. CAB stratified a higher percentage of the cohort (82%) as low-risk than IHC4 score (41.6%) and could re-stratify >74% of high Ki-67 and IHC4 score intermediate-risk zone patients into low-risk category. Overall the data suggest that CAB can effectively predict risk of distant recurrence with clear dichotomous high- or low-risk categorization.

Dataset Information

Weakly supervised temporal model for prediction of breast cancer distant recurrence.

Publications

Weakly supervised temporal model for prediction of breast cancer distant recurrence.

Similar Datasets

OmicsDI is part of the ELIXIR infrastructure

Tweets