What does this research mean for the field?

Extracting Social Determinants of Health (SDoH) from Chinese electronic medical records is feasible and covers all primary SDoH categories, though current data completeness is low and requires advanced natural language processing for effective utilization. Novelty: ClaimNovelty.METHODOLOGICAL. Consensus alignment: ConsensusAlignment.ESTABLISHES_NEW_DIRECTION.

What question did this study set out to answer?

To establish a China-specific classification framework and assess the feasibility, completeness, and availability of extracting social determinants of health from electronic medical records across healthcare institutions.

September 2, 2026Health care scienceOpen Access

Feasibility Assessment of Extracting Social Determinants of Health From Electronic Medical Records: A Multicenter Study Across Chinese Healthcare Institutions

View Full Paper

Ask AI

Bookmark

Authors

MGMengchun Gong ZOZihao Ouyang DMDandan Ma

Discussion

Member takes

Overview

Multicenter study reveals moderate availability but low completeness of social determinants of health across Chinese hospital records, highlighting the need for advanced data extraction tools.

Key Points

To establish a China-specific classification framework and assess the feasibility, completeness, and availability of extracting social determinants of health from electronic medical records across healthcare institutions.
Constructed a standardized framework of 50 social determinants of health features tailored to Chinese clinical care using literature synthesis and manual annotation of 2,000 electronic medical records.
Evaluated data completeness and availability across 43 electronic health record fields within over 5.6 million patient records from 40 hospitals in the National Clinical Research Data Center using a three-level hierarchical scoring system (maximum score of 3).
Electronic medical record fields covered all six primary social determinant categories and 25 of 50 specific attributes, but only seven features were fully extractable, with social and community context data largely missing.
Data availability achieved a mean score of 2.14 (95% CI: 1.83–2.44) out of 3, whereas composite completeness scored significantly lower with a mean of 1.47 (95% CI: 1.20–1.73) out of 3.

Cite This Study

Gong et al. (2026) studied this question.

synapsesocial.com/papers/6a97e2c5c562ede874ec70ad https://doi.org/10.1002/hcs2.70094

View Full Paper

Ask AI

Bookmark

September 2, 2026Health care scienceOpen Access

Feasibility Assessment of Extracting Social Determinants of Health From Electronic Medical Records: A Multicenter Study Across Chinese Healthcare Institutions

View Full Paper

Ask AI

Bookmark

Authors

MGMengchun Gong ZOZihao Ouyang DMDandan Ma

Discussion

Member takes

Overview

Multicenter study reveals moderate availability but low completeness of social determinants of health across Chinese hospital records, highlighting the need for advanced data extraction tools.

Key Points

To establish a China-specific classification framework and assess the feasibility, completeness, and availability of extracting social determinants of health from electronic medical records across healthcare institutions.
Constructed a standardized framework of 50 social determinants of health features tailored to Chinese clinical care using literature synthesis and manual annotation of 2,000 electronic medical records.
Evaluated data completeness and availability across 43 electronic health record fields within over 5.6 million patient records from 40 hospitals in the National Clinical Research Data Center using a three-level hierarchical scoring system (maximum score of 3).
Electronic medical record fields covered all six primary social determinant categories and 25 of 50 specific attributes, but only seven features were fully extractable, with social and community context data largely missing.
Data availability achieved a mean score of 2.14 (95% CI: 1.83–2.44) out of 3, whereas composite completeness scored significantly lower with a mean of 1.47 (95% CI: 1.20–1.73) out of 3.

Cite This Study

Gong et al. (2026) studied this question.

synapsesocial.com/papers/6a97e2c5c562ede874ec70ad https://doi.org/10.1002/hcs2.70094

View Full Paper

Ask AI

Bookmark