Diversity and Inclusion Within Datasets in Heart Failure

Q: What is the clinical evidence from this study?

Study design: Systematic Review. Population: Heart Failure (n=2000000). Intervention: Artificial intelligence datasets. Primary outcome: Dataset accessibility and reporting of demographic attributes.

Q: What does this research mean for the field?

Datasets used for developing artificial intelligence in heart failure lack consistent reporting of demographic attributes such as race, ethnicity, and socioeconomic status, and are predominantly composed of White/Caucasian individuals, highlighting a critical need for more diverse, transparent, and accessible data. Novelty: ClaimNovelty.SYNTHESIS. Consensus alignment: ConsensusAlignment.NEUTRAL.

Elinor Laws; Maria Charalambides; Sonam Vadera; Eva Keller; Joseph Alderman; Breanna Blackboro; Henry David Jeffry Hogg; Thomas S. Salisbury; Joanne Palmer; Melanie Calvert; Maxine Mackintosh; Rubeta Matin; Elizabeth Sapey; Johan Ordish; Melissa D. McCradden; Bilal A. Mateen; Jacqui Gath; Ade Adebajo; Stephanie Kuku; William Bradlow; Alastair K. Denniston; Xiaoxuan Liu

doi:10.1016/j.jacadv.2025.101610

Key result

Only 28% of HF AI datasets are fully accessible, with inconsistent demographic reporting.

n=2,000,000

Figures

Graphical abstract published by JACC Advances — Graphical abstract · Published by JACC Advances · CC BY-NC-ND

Why the study?

AI performance depends on whether training data reflects target populations, but the diversity and inclusivity of datasets used across the AI lifecycle in heart failure were not well characterized.

Study Design

Type

Systematic Review (n=2,000,000)

Structured PICO

P

Population

72 datasets representing over 2 million individuals from 23 countries used for developing AI health technologies related to heart failure

O

Outcome

Dataset accessibility, geographical origin, relevant metadata reporting (age, sex/gender, race/ethnicity, socioeconomic status), and dataset composition

Limitations

Search strategy is unlikely to be exhaustive of all available datasets.
Newer heart failure datasets published after August 2022 are unlikely to be included.
A large number of datasets were effectively inaccessible, limiting the ability to characterize their composition and metadata.
There may be additional embedded metadata within the data files or additional dataset documentation that could not be accessed.
Search strategy unlikely to be exhaustive of all available datasets
Newer HF datasets published after August 2022 unlikely to be included
Large number of datasets were effectively inaccessible, limiting ability to characterize them

Ask about this paper

Answers are grounded in this paper and cite their sources.

Implication

There is inconsistent reporting of demographic metadata and limited accessibility in datasets used to develop AI for heart failure, highlighting the need for transparent and inclusive data practices to ensure equitable AI technologies.

Abstract

BACKGROUND: Heart failure (HF) is a life-threatening disease affecting 64 million people worldwide. Artificial intelligence (AI) technologies are being developed for use in HF to support early diagnosis and stratification of treatment. The performance characteristics of AI technologies are influenced by whether the data used during the AI lifecycle reflects the populations for which the AI is used. OBJECTIVES: The aim of the study was to identify and characterize datasets used across the lifecycle of AI technologies for HF, focusing on data diversity and inclusivity. METHODS: MEDLINE and Embase were systematically searched from January 1, 2012, until August 30, 2022, to identify articles relating to the development of AI in HF. Articles were independently screened by 2 reviewers to identify datasets. Dataset documentation was analyzed with a focus on accessibility, geographical origin, relevant metadata reporting, and dataset composition. RESULTS: The 72 datasets identified represented 23 countries and over 2 million individuals. In total, 62 (86%) datasets reported "age," 61 (85%) reported sex or gender, 21 (29%) reported race and/or ethnicity, and 8 (11%) reported socioeconomic status. In the 21 datasets that reported race and/or ethnicity, 89% of individuals represented were reported within the "White" or "Caucasian" category. Only 20 (28%) datasets were fully accessible. CONCLUSIONS: Reporting of sex, gender, and socioeconomic status in HF datasets is inconsistent. There is a need to generate datasets that are transparently reported and accessible. Although collecting and reporting demographic attributes is complex and needs to be undertaken with appropriate safeguards, it is also an essential step toward building equitable AI-based health technologies.

Discussion

No takes yet. Share an insight, caveat, or question.

Cite This Study

Laws et al. (2025) conducted a systematic review in Heart Failure (n=2,000,000). Artificial intelligence datasets was evaluated on Dataset accessibility and reporting of demographic attributes. Among 72 datasets used for developing AI in heart failure, only 28% were fully accessible, and reporting of demographic attributes like race, ethnicity, and socioeconomic status was inconsistent.

synapsesocial.com/papers/6a1c23c126cb5670aa9d6da3 https://doi.org/10.1016/j.jacadv.2025.101610

View Full Paper

Ask AI

Bookmark

Share

Diversity and Inclusion Within Datasets in Heart Failure

Figures

Study Design

Structured PICO

Limitations

Ask about this paper

Implication

Abstract

Discussion

Cite This Study

Diversity and Inclusion Within Datasets in Heart Failure

Figures

Study Design

Structured PICO

Limitations

Ask about this paper

Implication

Abstract

Discussion

Cite This Study

Keep reading

Keep reading