In tasks involving the interpretation of medical images, suitably trained machine-learning models often exceed the performance of medical experts. Yet such a high-level of performance typically requires that the models be trained with relevant datasets that have been painstakingly annotated by experts. Here we show that a self-supervised model trained on chest X-ray images that lack explicit annotations performs pathology-classification tasks with accuracies comparable to those of radiologists. On an external validation dataset of chest X-rays, the self-supervised model outperformed a fully supervised model in the detection of three pathologies (out of eight), and the performance generalized to pathologies that were not explicitly annotated for model training, to multiple image-interpretation tasks and to datasets from multiple institutions. 

https://www.nature.com/articles/s41551-022-00936-9.pdf

Expert-level detection of pathologies from unannotated chest X-ray images via self-supervised learning

Design and Usability of Digital Libraries: Case Studies in the Asia Pacific

Purpose Next-generation sequencing (NGS) can facilitate precision medicine approaches in metastatic colorectal cancer (mCRC) patients. We investigated the molecular profiling of Korean mCRC patients under the K-MASTER project which was initiated in June 2017 as a nationwide precision medicine oncology clinical trial platform which used NGS assay to screen actionable mutations. Materials and methods As of 22 January 2020, total of 994 mCRC patients were registered in K-MASTER project. Targeted sequencing was performed using three platforms which were composed of the K-MASTER cancer panel v1.1 and the SNUH FIRST Cancer Panel v3.01. If tumor tissue was not available, cell-free DNA was extracted and the targeted sequencing was performed by Axen Cancer Panel as a liquid biopsy. Results In 994 mCRC patients, we found 1,564 clinically meaningful pathogenic variants which mutated in 71 genes. Anti-EGFR therapy candidates were 467 patients (47.0%) and BRAF V600E mutation (n=47, 4.7%), deficient mismatch repair/microsatellite instability-high (n=15, 1.5%), HER2 amplifications (n=10, 1.0%) could be incorporated with recently approved drugs. The patients with high tumor mutation burden (n=101, 12.7%) and DNA damaging response and repair defect pathway alteration (n=42, 4.2%) could be enrolled clinical trials with immune checkpoint inhibitors. There were more colorectal cancer molecular alterations such as PIK3CA, KRAS G12C, atypical BRAF, and HER2 mutations and even rarer but actionable genes that approved or ongoing clinical trials in other solid tumors. Conclusion K-MASTER project provides an intriguing background to investigate new clinical trials with biomarkers and give therapeutic opportunity for mCRC patients.

/pdf/clinical-application-of-targeted-deep-sequencing-in-36ch5gq4hi.pdf

Clinical Application of Targeted Deep Sequencing in Metastatic Colorectal Cancer Patients: Actionable Genomic Alteration in K-MASTER Project

Pathology reports primarily consist of unstructured free text and thus the clinical information contained in the reports is not trivial to access or query. Multiple natural language processing (NLP) techniques have been proposed to automate the coding of pathology reports via text classification. In this systematic review, we follow the guidelines proposed by the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA; Page et al., 2020: BMJ.) to identify the NLP systems for classifying pathology reports published between the years of 2010 and 2021. Based on our search criteria, a total of 3445 records were retrieved, and 25 articles met the final review criteria. We benchmarked the systems based on methodology, complexity of the prediction task and core types of NLP models: i) Rule-based and Intelligent systems, ii) statistical machine learning, and iii) deep learning. While certain tasks are well addressed by these models, many others have limitations and remain as open challenges, such as, extraction of many cancer characteristics (size, shape, type of cancer, others) from pathology reports. We investigated the final set of papers (25) and addressed their potential as well as their limitations. We hope that this systematic review helps researchers prioritize the development of innovated approaches to tackle the current limitations and help the advancement of cancer research. 

Automatic Classification of Cancer Pathology Reports: A Systematic Review

The current generation of sequencing technologies has led to significant advances in identifying novel disease-associated mutations and generated large amounts of data in a high-throughput manner. Such data in conjunction with clinical routine data are proven to be highly useful in deriving population-level and patient-level predictions, especially in the field of cancer precision medicine. However, data harmonization across multiple national and international clinical sites is an essential step for the assessment of events and outcomes associated with patients, which is currently not adequately addressed. The Observational Medical Outcomes Partnership (OMOP) Common Data Model (CDM) is an internationally established research data repository introduced by the Observational Health Data Science and Informatics (OHDSI) community to overcome this issue. To address the needs of cancer research, the genomic vocabulary extension was introduced in 2020 to support the standardization of subsequent data analysis. In this review, we evaluate the current potential of the OMOP CDM to be applicable in cancer prediction and how comprehensively the genomic vocabulary extension of the OMOP can serve current needs of AI-based predictions. For this, we systematically screened the literature for articles that use the OMOP CDM in predictive analyses in cancer and investigated the underlying predictive models/tools. Interestingly, we found 248 articles, of which most use the OMOP for harmonizing their data, but only 5 make use of predictive algorithms on OMOP-based data and fulfill our criteria. The studies present multicentric investigations, in which the OMOP played an essential role in discovering and optimizing machine learning (ML)-based models. Ultimately, the use of the OMOP CDM leads to standardized data-driven studies for multiple clinical sites and enables a more solid basis utilizing, e.g., ML models that can be reused and combined in early prediction, diagnosis, and improvement of personalized cancer care and biomarker discovery.

/pdf/omop-cdm-can-facilitate-data-driven-studies-for-cancer-t1a2wbae.pdf

OMOP CDM Can Facilitate Data-Driven Studies for Cancer Prediction: A Systematic Review

Pathology reports contain the essential data for both clinical and research purposes. However, the extraction of meaningful, qualitative data from the original document is difficult due to the narrative and complex nature of such reports. Keyword extraction for pathology reports is necessary to summarize the informative text and reduce intensive time consumption. In this study, we employed a deep learning model for the natural language process to extract keywords from pathology reports and presented the supervised keyword extraction algorithm. We considered three types of pathological keywords, namely specimen, procedure, and pathology types. We compared the performance of the present algorithm with the conventional keyword extraction methods on the 3115 pathology reports that were manually labeled by professional pathologists. Additionally, we applied the present algorithm to 36,014 unlabeled pathology reports and analysed the extracted keywords with biomedical vocabulary sets. The results demonstrated the suitability of our model for practical application in extracting important data from pathology reports.

/pdf/validation-of-deep-learning-natural-language-processing-2h3357d7y5.pdf

Validation of deep learning natural language processing algorithm for keyword extraction from pathology reports in electronic health records.

Purpose The purpose of this study was to identify the concordant or discordant genomic profiling between primary and matched metastatic tumors in patients with colorectal cancer (CRC) and to explore the clinical implication. Materials and methods Surgical samples of primary and matched metastatic tissues from 158 patients (335 samples) with CRC at Korea University Anam Hospital were evaluated using the Ion AmpliSeq Cancer Hotspot Panel. We compared genetic variants and classified them as concordant, primary-specific, and metastasis-specific variants. We used a combination of principal components analysis and clustering to find genomic groups. Kaplan-Meier curves were used to appraise survival between genomic groups. We used machine learning to confirm the correlation between genetic variants and metastatic sites. Results A total of 282 types of deleterious non-synonymous variants were selected for analysis. Of a total of 897 variants, an average of 40% was discordant. Three genomic groups were yielded based on the genomic discrepancy patterns. Overall survival differed significantly between the genomic groups. The poorest group had the highest proportion of concordant KRAS G12V and additional metastasis-specific SMAD4. Correlation analysis between genetic variants and metastatic sites suggested that concordant KRAS mutations would have more disseminated metastases. Conclusion Driver gene mutations were mostly concordant; however, discordant or metastasis-specific mutations were present. Clinically, the concordant driver genetic changes with additional metastasis-specific variants can predict poor prognosis for patients with CRC.

/pdf/clinical-implication-of-concordant-or-discordant-genomic-3m6br5hvmz.pdf

Clinical Implication of Concordant or Discordant Genomic Profiling between Primary and Matched Metastatic Tissues in Patients with Colorectal Cancer

Using Natural Language Process (NLP) as an efficient way to research paper is important when user feedback is sparse or unavailable. The task of text mining research paper is challenging, mainly due to the problem of unique characteristics such as jargon. Nowadays, there exist many language models that learn deep semantic representations by being trained on huge corpora. In this paper, we specify the NLP pre-processing process with Economics journal paper and apply it to a deep learning model to extract keywords. Here, we focus on the strength of NLP when applied to an unknown field. The analysis result shows the possibility and potential usefulness of the relationship research between keywords in research papers.

Keyword Extraction in Economics Literatures using Natural Language Processing

/pdf/korean-validation-of-the-short-version-of-the-temps-a-2q6yv8ae.pdf

Korean Validation of the Short Version of the TEMPS-A (Temperament Evaluation of Memphis, Pisa, Paris, and San Diego Autoquestionnaire) in Patients with Mood Disorders

Pathology is a basic medical field of diagnosing diseases and describing their conditions that are presented in a medical report form. It is important to understand the medical domain from various medical terms in the reports, such as observation parts, disease conditions and names, and measuring units. In this paper, we apply various machine learning algorithms to predict the domain of untrained terms used in real pathological reports. Here, we focus on the oncology section with ICD-O3. The analysis result shows the possibility and potential usefulness of the domain prediction using medical terms.

Sunho Choi

Papers

Validation of deep learning natural language processing algorithm for keyword extraction from pathology reports in electronic health records.

Clinical Implication of Concordant or Discordant Genomic Profiling between Primary and Matched Metastatic Tissues in Patients with Colorectal Cancer

Keyword Extraction in Economics Literatures using Natural Language Processing

Korean Validation of the Short Version of the TEMPS-A (Temperament Evaluation of Memphis, Pisa, Paris, and San Diego Autoquestionnaire) in Patients with Mood Disorders

Vocabulary Domain Prediction for Pathological Report Analysis Using ICD-O3