Systematic Review of Large Language Model Limitations in Healthcare

A systematic review of 81 studies published between 2018 and 2025 identifies limitations of large language models in providing healthcare content. The dataset, created by researcher Mohsen Khosravi and last updated in April 2026, contains the analysis framework and results from this review. It includes themes like data limitations, input quality dependence, and ethical concerns derived from the included studies.

Use Cases

• Analyze the frequency of cited limitations like 'accuracy gap' or 'data limitations' across the 81 reviewed studies.
• Apply the Input-Process-Output (IPO) model framework to categorize identified themes from the review data.
• Correlate study quality scores from the AACODS checklist with the types of limitations reported.
• Extract and compare key themes such as 'model design constraints' or 'ethical concerns' for meta-analysis.

Strengths

• Analysis based on 81 included studies from major databases like PubMed and Scopus.
• Studies were assessed for quality using the standardized AACODS checklist.
• Clear temporal coverage of literature from 2018 to 2025.

Limitations

• Dataset is very small at 5.5 KB, indicating limited raw data or summary-level information only.
• Scope is restricted to English-language publications, introducing a language bias.
• Thematic analysis results are presented, but the underlying extracted data from each study may not be fully included.

Provenance

Source

Systematic review conducted by researcher Mohsen Khosravi, shared on figshare.

Collection Method

Literature search in PubMed, Scopus, and Cochrane Database, followed by screening, quality assessment (AACODS), and thematic analysis.

Time Range

Studies published between 2018 and 2025, review conducted in September 2025.

Freshness

Last updated in April 2026, based on a review conducted in September 2025.