RESUMO
BACKGROUND: Cancer recurrence is an important measure of the impact of cancer treatment. However, no population-based data on recurrence are available. Pathology reports could potentially identify cancer recurrences. Their utility to capture recurrences is unknown. OBJECTIVE: This analysis assesses the sensitivity of pathology reports to identify patients with cancer recurrence and the stage at recurrence. SUBJECTS: The study includes patients with recurrent breast (n=214) or colorectal (n=203) cancers. RESEARCH DESIGN: This retrospective analysis included patients from a population-based cancer registry who were part of the Patient-Centered Outcomes Research (PCOR) Study, a project that followed cancer patients in-depth for 5 years after diagnosis to identify recurrences. MEASURES: Information abstracted from pathology reports for patients with recurrence was compared with their PCOR data (gold standard) to determine what percent had a pathology report at the time of recurrence, the sensitivity of text in the report to identify recurrence, and if the stage at recurrence could be determined from the pathology report. RESULTS: One half of cancer patients had a pathology report near the time of recurrence. For patients with a pathology report, the report's sensitivity to identify recurrence was 98.1% for breast cancer cases and 95.7% for colorectal cancer cases. The specific stage at recurrence from the pathology report had a moderate agreement with gold-standard data. CONCLUSIONS: Pathology reports alone cannot measure population-based recurrence of solid cancers but can identify specific cohorts of recurrent cancer patients. As electronic submission of pathology reports increases, these reports may identify specific recurrent patients in near real-time.
Assuntos
Documentação/normas , Neoplasias/diagnóstico , Neoplasias/patologia , Recidiva , Neoplasias da Mama/diagnóstico , Neoplasias da Mama/epidemiologia , Neoplasias da Mama/patologia , Neoplasias Colorretais/diagnóstico , Neoplasias Colorretais/epidemiologia , Neoplasias Colorretais/patologia , Documentação/métodos , Documentação/estatística & dados numéricos , Feminino , Humanos , Masculino , Pessoa de Meia-Idade , Neoplasias/epidemiologia , Estudos RetrospectivosRESUMO
The coronavirus disease 2019 (COVID-19) pandemic led to delayed medical care in the United States. We examined changes in patterns of cancer diagnosis and surgical treatment between January 1 and December 31 in 2020 and 2019 with real-time electronic pathology report data from population-based Surveillance, Epidemiology, and End Results cancer registries from Georgia and Louisiana. During 2020, there were 29 905 fewer pathology reports than in 2019, representing a 10.2% decline. Declines were observed in all age groups, including children and adolescents younger than 18 years. The nadir was early April 2020, with 42.8% fewer reports than in April 2019. Numbers of reports through December 2020 never consistently exceeded those in 2019 after first declines. Patterns were similar by age group and cancer site. Findings suggest substantial delays in diagnosis and treatment services for cancers during the pandemic. Ongoing evaluation can inform public health efforts to minimize any lasting adverse effects of the pandemic on cancer diagnosis, stage, treatment, and survival.
Assuntos
COVID-19 , Neoplasias , Adolescente , COVID-19/epidemiologia , Criança , Humanos , Neoplasias/diagnóstico , Neoplasias/epidemiologia , Neoplasias/terapia , Pandemias , Vigilância da População , Sistema de Registros , Estados Unidos/epidemiologiaRESUMO
Population cancer registries can benefit from Deep Learning (DL) to automatically extract cancer characteristics from the high volume of unstructured pathology text reports they process annually. The success of DL to tackle this and other real-world problems is proportional to the availability of large labeled datasets for model training. Although collaboration among cancer registries is essential to fully exploit the promise of DL, privacy and confidentiality concerns are main obstacles for data sharing across cancer registries. Moreover, DL for natural language processing (NLP) requires sharing a vocabulary dictionary for the embedding layer which may contain patient identifiers. Thus, even distributing the trained models across cancer registries causes a privacy violation issue. In this paper, we propose DL NLP model distribution via privacy-preserving transfer learning approaches without sharing sensitive data. These approaches are used to distribute a multitask convolutional neural network (MT-CNN) NLP model among cancer registries. The model is trained to extract six key cancer characteristics - tumor site, subsite, laterality, behavior, histology, and grade - from cancer pathology reports. Using 410,064 pathology documents from two cancer registries, we compare our proposed approach to conventional transfer learning without privacy-preserving, single-registry models, and a model trained on centrally hosted data. The results show that transfer learning approaches including data sharing and model distribution outperform significantly the single-registry model. In addition, the best performing privacy-preserving model distribution approach achieves statistically indistinguishable average micro- and macro-F1 scores across all extraction tasks (0.823,0.580) as compared to the centralized model (0.827,0.585).
RESUMO
OBJECTIVE: We implement 2 different multitask learning (MTL) techniques, hard parameter sharing and cross-stitch, to train a word-level convolutional neural network (CNN) specifically designed for automatic extraction of cancer data from unstructured text in pathology reports. We show the importance of learning related information extraction (IE) tasks leveraging shared representations across the tasks to achieve state-of-the-art performance in classification accuracy and computational efficiency. MATERIALS AND METHODS: Multitask CNN (MTCNN) attempts to tackle document information extraction by learning to extract multiple key cancer characteristics simultaneously. We trained our MTCNN to perform 5 information extraction tasks: (1) primary cancer site (65 classes), (2) laterality (4 classes), (3) behavior (3 classes), (4) histological type (63 classes), and (5) histological grade (5 classes). We evaluated the performance on a corpus of 95 231 pathology documents (71 223 unique tumors) obtained from the Louisiana Tumor Registry. We compared the performance of the MTCNN models against single-task CNN models and 2 traditional machine learning approaches, namely support vector machine (SVM) and random forest classifier (RFC). RESULTS: MTCNNs offered superior performance across all 5 tasks in terms of classification accuracy as compared with the other machine learning models. Based on retrospective evaluation, the hard parameter sharing and cross-stitch MTCNN models correctly classified 59.04% and 57.93% of the pathology reports respectively across all 5 tasks. The baseline models achieved 53.68% (CNN), 46.37% (RFC), and 36.75% (SVM). Based on prospective evaluation, the percentages of correctly classified cases across the 5 tasks were 60.11% (hard parameter sharing), 58.13% (cross-stitch), 51.30% (single-task CNN), 42.07% (RFC), and 35.16% (SVM). Moreover, hard parameter sharing MTCNNs outperformed the other models in computational efficiency by using about the same number of trainable parameters as a single-task CNN. CONCLUSIONS: The hard parameter sharing MTCNN offers superior classification accuracy for automated coding support of pathology documents across a wide range of cancers and multiple information extraction tasks while maintaining similar training and inference time as those of a single task-specific model.
Assuntos
Armazenamento e Recuperação da Informação/métodos , Aprendizado de Máquina , Processamento de Linguagem Natural , Neoplasias/patologia , Redes Neurais de Computação , Sistema de Registros , Humanos , Neoplasias/classificação , Máquina de Vetores de SuporteRESUMO
BACKGROUND: Since 2012, the Lower Anogenital Squamous Terminology (LAST) Project recommended a 2-tiered nomenclature, low-grade and high-grade squamous intraepithelial lesion (LSIL and HSIL), to replace the 3-tiered cervical intraepithelial neoplasia (CIN) system for HPV-associated lesions. Prior to 2019, preinvasive cervical lesions classified as CIN3, severe dysplasia, carcinoma in situ (CIS), and adenocarcinoma in situ (AIS) were considered reportable to the Louisiana Tumor Registry for a CIN3 project funded by the Centers for Disease Control and Prevention (CDC); but lesions classified exclusively as high-grade/HSIL based on the 2-tiered system were not considered reportable. Due to the terminology changes, we wanted to know whether pre-2019 reportable criteria need to be modified to capture all reportable precancerous cervical cases diagnosed in 2019 forward. OBJECTIVES: To evaluate the utilization of LAST 2-tiered classification, low-grade and high-grade squamous intraepithelial lesion, and p16 immunohistochemistry (IHC) testing on cervical biopsy/surgical specimens, assess the search criteria needed to identify high-grade lesions for the CDC-funded CIN3 project, and assess the impact of underreporting cervical lesions caused by terminology changes. METHODS: An equal number of abnormal/precancerous and normal cervical findings from biopsy pathology reports received in 2015 were randomly selected by an artificial intelligence (AI) search engine developed by Artificial Intelligence in Medicine Inc (AIM) using pre2019 search criteria. Selected pathology reports were reflagged for the reportability by AIM audit software based on 2019 search criteria and manually reviewed for the use of reportable terms including CIN3, severe dysplasia, CIS, AIS, highgrade/HSIL terminology, and CIN2 or CIN2-3 with positive p16 IHC testing. Cohen's kappa statistic was used to assess the agreement between AIM auto-coding and manual review. Positive predictive values (PPV) and sensitivity tests were computed to evaluate the reportable terms. RESULTS: Six out of 9 surveyed laboratories used 2-tiered terminology on cervical biopsy pathology reports and 7 performed p16 IHC tests. Of 1,974 randomly selected reports from 5 laboratories, 987 were flagged as precancer by AI using pre-2019 search criteria. After adding the high-grade/HSIL term into pre-2019 search criteria, precancerous reports increased by 29%. After manual review, 41.6% of these cases were reportable precancerous cervical cases with a PPV of 0.65 (95% CI, 0.62-0.67) and 13.6% had p16 IHC performed. CONCLUSIONS: Both the 2-tiered and 3-tiered nomenclature are needed to ensure complete identification of all reportable high-grade cervical lesions.
RESUMO
BACKGROUND: In order to comply with the Louisiana legislative obligation and meet funding agencies' requirement of case completeness for 12-month data submission, hospital cancer registries are mandated to submit cancer incidence data to the Louisiana Tumor Registry (LTR) within 6 months of diagnosis. However, enforcing compliance with timely reporting may result in incomplete data on adjuvant treatment received by the LTR. Although additional treatment information can be obtained via retransmission of the North American Association of Central Cancer Registries (NAACCR)modified abstracts, consolidating multiple NAACCR-modified abstracts for the same case is extremely time consuming. To avoid a huge amount of work while obtaining timely and complete data, the LTR has requested hospital cancer registries resubmit their data 15 months after the close of the diagnosis year. The purpose of this report is to assess the improvement in the completeness of data items related to treatment, staging and site specific factors. METHODS: The LTR requested that hospital cancer registries resubmit 15-month data between April 1, 2016 and April 15, 2016 for cases diagnosed in 2014. Microsoft Visual Studio Visual Basic script was used to link and compare resubmitted data with existing data in the LTR database. Data elements used for matching same patient/tumor were name, Social Security number, date of birth, primary site, laterality, and hospital identifier number. Treatment data items were compared as known vs none/ unknown and known vs known with different code. Matched records with updated information were imported into the LTR database and flagged as modified abstract records for manual consolidation. Nonmatched records were also loaded in the LTR database as potential new cases for further investigation. RESULTS: A total of 25,207 resubmitted NAACCR abstracts were received from 38 hospitals and freestanding radiation centers. About 11.1% had at least 1 update related to treatment and/or other data item; an average of 3.3 updates per updated abstract. The majority of the updates (45.7%) for treatment were changes from none/unknown to known value and 42.6% of the updates were related to radiation treatment fields. In addition, 172 potential new cases were identified. Approximately 10.5% (18 cases) of these new cases were confirmed to be truly missed cases after investigation. CONCLUSION: The 15-month data resubmission is a cost-effective approach to obtaining complete information on treatment and other key data items from reporting facilities and can also be used to identify potential missed cases.