Dataset Information

Artificial intelligence-aided clinical annotation of a large multi-cancer genomic dataset.

ABSTRACT: To accelerate cancer research that correlates biomarkers with clinical endpoints, methods are needed to ascertain outcomes from electronic health records at scale. Here, we train deep natural language processing (NLP) models to extract outcomes for participants with any of 7 solid tumors in a precision oncology study. Outcomes are extracted from 305,151 imaging reports for 13,130 patients and 233,517 oncologist notes for 13,511 patients, including patients with 6 additional cancer types. NLP models recapitulate outcome annotation from these documents, including the presence of cancer, progression/worsening, response/improvement, and metastases, with excellent discrimination (AUROC > 0.90). Models generalize to cancers excluded from training and yield outcomes correlated with survival. Among patients receiving checkpoint inhibitors, we confirm that high tumor mutation burden is associated with superior progression-free survival ascertained using NLP. Here, we show that deep NLP can accelerate annotation of molecular cancer datasets with clinically meaningful endpoints to facilitate discovery.

SUBMITTER: Kehl KL

PROVIDER: S-EPMC8674229 | biostudies-literature | 2021 Dec

REPOSITORIES: biostudies-literature

ACCESS DATA

Publications

Artificial intelligence-aided clinical annotation of a large multi-cancer genomic dataset.

Kehl Kenneth L KL Xu Wenxin W Gusev Alexander A Bakouny Ziad Z Choueiri Toni K TK Riaz Irbaz Bin IB Elmarakeby Haitham H Van Allen Eliezer M EM Schrag Deborah D

Nature communications 20211215 1

To accelerate cancer research that correlates biomarkers with clinical endpoints, methods are needed to ascertain outcomes from electronic health records at scale. Here, we train deep natural language processing (NLP) models to extract outcomes for participants with any of 7 solid tumors in a precision oncology study. Outcomes are extracted from 305,151 imaging reports for 13,130 patients and 233,517 oncologist notes for 13,511 patients, including patients with 6 additional cancer types. NLP mod ...[more]

PMID: 34911934

Similar Datasets

Project description:BackgroundThe early detection of high-grade prostate cancer (HGPCa) is of great importance. However, the current detection strategies result in a high rate of negative biopsies and high medical costs. In this study, we aimed to establish an Asian Prostate Cancer Artificial intelligence (APCA) score with no extra cost other than routine health check-ups to predict the risk of HGPCa.Patients and methodsA total of 7476 patients with routine health checkup data who underwent prostate biopsies from January 2008 to December 2021 in eight referral centers in Asia were screened. After data pre-processing and cleaning, 5037 patients and 117 features were analyzed. Seven AI-based algorithms were tested for feature selection and seven AI-based algorithms were tested for classification, with the best combination applied for model construction. The APAC score was established in the CH cohort and validated in a multi-center cohort and in each validation cohort to evaluate its generalizability in different Asian regions. The performance of the models was evaluated using area under the receiver operating characteristic curve (ROC), calibration plot, and decision curve analyses.ResultsEighteen features were involved in the APCA score predicting HGPCa, with some of these markers not previously used in prostate cancer diagnosis. The area under the curve (AUC) was 0.76 (95% CI:0.74-0.78) in the multi-center validation cohort and the increment of AUC (APCA vs. PSA) was 0.16 (95% CI:0.13-0.20). The calibration plots yielded a high degree of coherence and the decision curve analysis yielded a higher net clinical benefit. Applying the APCA score could reduce unnecessary biopsies by 20.2% and 38.4%, at the risk of missing 5.0% and 10.0% of HGPCa cases in the multi-center validation cohort, respectively.ConclusionsThe APCA score based on routine health checkups could reduce unnecessary prostate biopsies without additional examinations in Asian populations. Further prospective population-based studies are warranted to confirm these results.

Project description:BackgroundTuberculosis (TB) remains a global public health challenge. The existing Bacillus Calmette-Guérin vaccine has limited efficacy in preventing adult pulmonary TB, necessitating the development of new vaccines with improved protective effects.MethodsComputer-aided design and artificial intelligence technologies, combined with bioinformatics and immunoinformatics approaches, were used to design a multi-epitope vaccine (MEV) against TB. Comprehensive bioinformatics analyses were conducted to evaluate the physicochemical properties, spatial structure, immunogenicity, molecular dynamics (MD), and immunological characteristics of the MEV.ResultsWe constructed a MEV, designated ZL12138L, containing 13 helper T lymphocyte epitopes, 12 cytotoxic T lymphocyte epitopes, 8 B-cell epitopes, as well as Toll-like receptor (TLR) agonists and helper peptides. Bioinformatics analyses revealed that ZL12138L should exhibit excellent immunogenicity and antigenicity, with no toxicity or allergenicity, and had potential to induce robust immune responses and high solubility, the immunogenicity score was 4.14449, the antigenicity score was 0.8843, and the immunological score was 0.470. Moreover, ZL12138L showed high population coverage for human leukocyte antigen class I and II alleles, reaching 92.41% and 90.17%, respectively, globally. Molecular docking analysis indicated favorable binding affinity of ZL12138L with TLR-2 and TLR-4, with binding energies of -1173.4 and -1360.5 kcal/mol, respectively. Normal mode analysis and MD simulations indicated the stability and dynamic properties of the vaccine construct. Immune simulation predictions suggested that ZL12138L could effectively activate innate and adaptive immune cells, inducing high levels of Type 1 T helper cell cytokines.ConclusionsThis study provides compelling evidence for ZL12138L as a promising TB vaccine candidate. Future research will focus on experimental validation and further optimization of the vaccine design.

Project description:BackgroundWe investigated the performance improvement of physicians with varying levels of chest radiology experience when using a commercially available artificial intelligence (AI)-based computer-assisted detection (CAD) software to detect lung cancer nodules on chest radiographs from multiple vendors.MethodsChest radiographs and their corresponding chest CT were retrospectively collected from one institution between July 2017 and June 2018. Two author radiologists annotated pathologically proven lung cancer nodules on the chest radiographs while referencing CT. Eighteen readers (nine general physicians and nine radiologists) from nine institutions interpreted the chest radiographs. The readers interpreted the radiographs alone and then reinterpreted them referencing the CAD output. Suspected nodules were enclosed with a bounding box. These bounding boxes were judged correct if there was significant overlap with the ground truth, specifically, if the intersection over union was 0.3 or higher. The sensitivity, specificity, accuracy, PPV, and NPV of the readers' assessments were calculated.ResultsIn total, 312 chest radiographs were collected as a test dataset, including 59 malignant images (59 nodules of lung cancer) and 253 normal images. The model provided a modest boost to the reader's sensitivity, particularly helping general physicians. The performance of general physicians was improved from 0.47 to 0.60 for sensitivity, from 0.96 to 0.97 for specificity, from 0.87 to 0.90 for accuracy, from 0.75 to 0.82 for PPV, and from 0.89 to 0.91 for NPV while the performance of radiologists was improved from 0.51 to 0.60 for sensitivity, from 0.96 to 0.96 for specificity, from 0.87 to 0.90 for accuracy, from 0.76 to 0.80 for PPV, and from 0.89 to 0.91 for NPV. The overall increase in the ratios of sensitivity, specificity, accuracy, PPV, and NPV were 1.22 (1.14-1.30), 1.00 (1.00-1.01), 1.03 (1.02-1.04), 1.07 (1.03-1.11), and 1.02 (1.01-1.03) by using the CAD, respectively.ConclusionThe AI-based CAD was able to improve the ability of physicians to detect nodules of lung cancer in chest radiographs. The use of a CAD model can indicate regions physicians may have overlooked during their initial assessment.

Dataset Information

Artificial intelligence-aided clinical annotation of a large multi-cancer genomic dataset.

Publications

Artificial intelligence-aided clinical annotation of a large multi-cancer genomic dataset.

Similar Datasets

OmicsDI is part of the ELIXIR infrastructure

Tweets