The Sociodemographic Biases in Machine Learning Algorithms: A Biomedical Informatics Perspective.

Artificial intelligence models represented in machine learning algorithms are promising tools for risk assessment used to guide clinical and other health care decisions. Machine learning algorithms, however, may house biases that propagate stereotypes, inequities, and discrimination that contribute to socioeconomic health care disparities. The biases include those related to some sociodemographic characteristics such as race, ethnicity, gender, age, insurance, and socioeconomic status from the use of erroneous electronic health record data. Additionally, there is concern that training data and algorithmic biases in large language models pose potential drawbacks. These biases affect the lives and livelihoods of a significant percentage of the population in the United States and globally. The social and economic consequences of the associated backlash cannot be underestimated. Here, we outline some of the sociodemographic, training data, and algorithmic biases that undermine sound health care risk assessment and medical decision-making that should be addressed in the health care system. We present a perspective and overview of these biases by gender, race, ethnicity, age, historically marginalized communities, algorithmic bias, biased evaluations, implicit bias, selection/sampling bias, socioeconomic status biases, biased data distributions, cultural biases and insurance status bias, conformation bias, information bias and anchoring biases and make recommendations to improve large language model training data, including de-biasing techniques such as counterfactual role-reversed sentences during knowledge distillation, fine-tuning, prefix attachment at training time, the use of toxicity classifiers, retrieval augmented generation and algorithmic modification to mitigate the biases moving forward.

CDS-Compare: A Web Application for Machine Learning Assisted Curation of Clinical Order Sets.

Order sets that adhere to disease-specific guidelines have been shown to increase clinician efficiency and patient safety but curating these order sets, particularly for consistency across multiple sites, is difficult and time consuming. We created software called CDS-Compare to alleviate the burden on expert reviewers in rapidly and effectively curating large databases of order sets. We applied our clustering-based software to a database of NLP-processed order sets extracted from VA's Electronic Health Record, then had subject-matter experts review the web application version of our software for clustering validity.

Automated Modeling of Clinical Narrative with High Definition Natural Language Processing Using Solor and Analysis Normal Form.

OBJECTIVE: One important concept in informatics is data which meets the principles of Findability, Accessibility, Interoperability and Reusability (FAIR). Standards, such as terminologies (findability), assist with important tasks like interoperability, Natural Language Processing (NLP) (accessibility) and decision support (reusability). One terminology, Solor, integrates SNOMED CT, LOINC and RxNorm. We describe Solor, HL7 Analysis Normal Form (ANF), and their use with the high definition natural language processing (HD-NLP) program. METHODS: We used HD-NLP to process 694 clinical narratives prior modeled by human experts into Solor and ANF. We compared HD-NLP output to the expert gold standard for 20% of the sample. Each clinical statement was judged "correct" if HD-NLP output matched ANF structure and Solor concepts, or "incorrect" if any ANF structure or Solor concepts were missing or incorrect. Judgements were summed to give totals for "correct" and "incorrect". RESULTS: 113 (80.7%) correct, 26 (18.6%) incorrect, and 1 error. Inter-rater reliability was 97.5% with Cohen's kappa of 0.948. CONCLUSION: The HD-NLP software provides useable complex standards-based representations for important clinical statements designed to drive CDS.