Automating the overburdened clinical coding system: challenges and next steps

Review Engineering, Biomedical

Self-supervised learning in medicine and healthcare

Rayan Krishnan et al.

Summary: This Review examines the advantages and limitations of self-supervised methods and models in medicine and healthcare, as well as the challenges in collecting unbiased data for their training. The utilization of self-supervised learning has the potential to accelerate the development of medical artificial intelligence, particularly in tasks involving multimodal datasets and the analysis of electronic health records, medical images, bioelectrical signals, and genetic and protein sequences and structures.

NATURE BIOMEDICAL ENGINEERING (2022)

Add to Collection

Article Computer Science, Interdisciplinary Applications

Note Bloat impacts deep learning-based NLP models for clinical prediction tasks

Jinghui Liu et al.

Summary: The implementation of Electronic Health Records (EHR) has led to the overuse of content-importing technology, resulting in bloated notes with redundant content. This study examines the impact of redundancy on deep learning-based NLP models for clinical prediction tasks. The findings suggest that removing redundancy usually has little negative impact on the model performance and can even improve results in certain cases. Additionally, it is shown that model predictions can be manipulated by adding note duplicates.

JOURNAL OF BIOMEDICAL INFORMATICS (2022)

Add to Collection

Article Health Care Sciences & Services

Automated clinical coding: what, why, and where we are?

Hang Dong et al.

Summary: Clinical coding is the process of transforming medical information into structured codes for statistical analysis. Automated clinical coding is a promising task for AI, but there are challenges in achieving explainability and consistency. Involvement of coders is necessary in the development process.

NPJ DIGITAL MEDICINE (2022)

Add to Collection

Article Computer Science, Information Systems

FairLens: Auditing black-box clinical decision support systems

Cecilia Panigutti et al.

Summary: The widespread use of algorithmic decision-making raises concerns about unintended bias in AI systems, especially in critical settings like healthcare. FairLens is introduced as a method to detect and explain biases in models, helping healthcare experts identify and address biases before using the model in clinical decision-making.

INFORMATION PROCESSING & MANAGEMENT (2021)

Add to Collection

Article Medical Informatics