A hybrid system to understand the relations between assessments and plans in progress notes

Article Computer Science, Artificial Intelligence

Automated segmentation of knee articular cartilage: Joint deep and hand-crafted learning-based framework using diffeomorphic mapping

Somayeh Ebrahimkhani et al.

Summary: The study proposed a joint deep and hand-crafted learning-based framework for automated segmentation of knee articular cartilage tissue, achieving high segmentation accuracy and volume correlation on multiple datasets.

NEUROCOMPUTING (2022)

添加到收藏夹

Article Engineering, Biomedical

Diagnosis of Parkinson's disease based on SHAP value feature selection

Yuchun Liu et al.

Summary: This paper introduces a feature selection method using SHAP values and applies it to the diagnosis and classification of Parkinson's disease. The experimental results demonstrate that the SHAP-value-based feature selection method achieves good classification performance in the diagnosis of Parkinson's disease.

BIOCYBERNETICS AND BIOMEDICAL ENGINEERING (2022)

添加到收藏夹

Article Automation & Control Systems

Med-BERT: A Pretraining Framework for Medical Records Named Entity Recognition

Ning Liu et al.

Summary: This article introduces the Med-BERT method for improving medical records NER tasks, showing good performance in experiments. The method provides better representations of long medical entities while protecting privacy information, and is essential for clinical decision support systems and medical real-world research.

IEEE TRANSACTIONS ON INDUSTRIAL INFORMATICS (2022)

添加到收藏夹

Article Computer Science, Artificial Intelligence

Multi-domain clinical natural language processing with MedCAT: The Medical Concept Annotation Toolkit

Zeljko Kraljevic et al.

Summary: MedCAT is an open source Medical Concept Annotation Toolkit that provides a novel self-supervised machine learning algorithm, an annotation interface, and integrations with the broader CogStack ecosystem. It shows improved performance in extracting UMLS concepts and strong transferability between hospitals, datasets, and concept types through real-world validation with self-supervised training and fine-tuning.

ARTIFICIAL INTELLIGENCE IN MEDICINE (2021)

添加到收藏夹

Article Oncology

Natural Language Processing to Identify Cancer Treatments With Electronic Medical Records

Jiaming Zeng et al.

Summary: By combining structured and unstructured data, we successfully identified the initial treatments administered to cancer patients, with the best results achieved using free-text clinical notes. This approach can be applied to treatment cohort identification and shows potential for longitudinal cancer treatment identification.

JCO CLINICAL CANCER INFORMATICS (2021)

添加到收藏夹

Article Computer Science, Artificial Intelligence

Compressing Large-Scale Transformer-Based Models: A Case Study on BERT

Prakhar Ganesh et al.

Summary: The research summarizes the study of compressing Transformer models, focusing on the BERT model. It investigates the latest compression techniques for BERT, clarifies the best practices for compressing large-scale Transformer models, and provides insights into the workings of various methods.

TRANSACTIONS OF THE ASSOCIATION FOR COMPUTATIONAL LINGUISTICS (2021)

添加到收藏夹

Article Computer Science, Artificial Intelligence

Explainable Artificial Intelligence for Human Decision Support System in the Medical Domain

Samanta Knapic et al.

Summary: This paper explores the potential of Explainable Artificial Intelligence methods for decision support in medical image analysis, aiming to improve the comprehensibility of decisions provided by Convolutional Neural Networks (CNNs) through various explainable methods. Results from human evaluation and user studies show that the Contextual Importance and Utility (CIU) method outperforms LIME and SHAP in enhancing human decision-making support, transparency, and user understanding.

MACHINE LEARNING AND KNOWLEDGE EXTRACTION (2021)

添加到收藏夹

Article Computer Science, Information Systems

Deep learning approaches for extracting adverse events and indications of dietary supplements from clinical text

Yadan Fan et al.

Summary: This study demonstrated the feasibility of using deep learning models to extract safety signals related to dietary supplements in clinical text. Deep learning models performed better than traditional models in named entity recognition and relation extraction tasks, showing great potential for monitoring the safety of dietary supplement use.

JOURNAL OF THE AMERICAN MEDICAL INFORMATICS ASSOCIATION (2021)

添加到收藏夹

Article Medical Informatics