Rubric10-Semantic Evaluation Report

Fields:

Rationale: id present (https://fairhub.io/datasets/2). DOI absent - should be 10.5281/zenodo.10642459 based on external_resources. RRID absent.

Fields:

Rationale: Both title and description present and comprehensive. Description includes study design, participant count, sites, data domains, and scientific rationale.

Fields:

Rationale: 17 keywords provided covering disease (T2DM), methodology (AI/ML, multimodal), study design (triple-balanced, cross-sectional), and data types (retinal imaging, wearables, biospecimens).

Fields:

Rationale: Page field present (https://fairhub.io/datasets/2). Resources field absent but external_resources provides extensive documentation links.

Fields:

Rationale: Neither parent_datasets nor related_datasets fields present. This is a standalone dataset but could reference Bridge2AI program datasets.

Fields:

Rationale: license_and_use_terms present and comprehensive with CC BY-NC 4.0 license details. ip_restrictions field absent but not critical for this public dataset.

Fields:

Rationale: Neither regulatory_restrictions nor confidentiality_level present despite having controlled access data with genetic sequencing and PHI.

Fields:

Rationale: download_url field absent but access_urls provided in distribution_formats (all pointing to https://fairhub.io/datasets/2).

Fields:

Rationale: distribution_formats comprehensively documented with 4 formats: DICOM (imaging), CSV (tabular/time-series), mHealth (wearables), REDCap (metadata). Top-level format/media_type absent.

Fields:

Rationale: external_resources extensively documented with 9 resources including project website, documentation, FAIRhub, Bridge2AI, NIH RePORTER, Zenodo, and publications. related_datasets absent.

Fields:

Rationale: CC BY-NC 4.0 license clearly allows reuse with attribution for non-commercial purposes. Controlled access terms mentioned but not detailed.

Fields:

Rationale: Top-level format and encoding fields absent. Formats documented in distribution_formats but not at dataset level.

Fields:

Rationale: Neither conforms_to nor conforms_to_schema present despite documentation referencing OMOP CDM, DICOM, mHealth standards.

Fields:

Rationale: variables field completely absent despite extensive multi-domain data collection documented in acquisition_methods.

Fields:

Rationale: intended_uses (5 items) and discouraged_uses (3 items) present. prohibited_uses absent.

Fields:

Rationale: N/A

Fields:

Rationale: human_subject_research extensively documented with IRB approvals from University of Washington (STUDY00016228) and reliance agreements from UAB and UCSD. ethical_reviews field absent but covered in human_subject_research.

Fields:

Rationale: is_deidentified field completely absent despite clear documentation of de-identification practices in subset descriptions.

Fields:

Rationale: participant_privacy field absent. Privacy mentioned in sensitive_elements but no structured privacy protections documented.

Fields:

Rationale: informed_consent mentioned in human_subject_research (written informed consent provided by all participants) but no separate informed_consent structured field.

Fields:

Rationale: vulnerable_populations mentioned in human_subject_research (special_populations includes racial/ethnic minorities, tribal consultation). participant_compensation field absent.

Fields:

Rationale: N/A

Fields:

Rationale: subpopulations comprehensively documented with 8 subpopulations: 4 race/ethnicity groups (Asian, Black, Hispanic, White) and 4 diabetes severity levels. Target distributions specified (25% each, ~1,000 participants).

Fields:

Rationale: instances documented with target 4,000 participants, inclusion/exclusion criteria, age requirement (40+), and instance_type specifying human participants from three sites.

Fields:

Rationale: Both variables and is_tabular fields absent. Multiple tabular data sources documented (surveys, labs, wearables) but no variable metadata or tabular flag.

Fields:

Rationale: Data topics clear from instances (T2DM participants) and throughout metadata. Medical condition (Type 2 Diabetes Mellitus) and research focus (salutogenesis) well-documented.

Fields:

Rationale: sampling_strategies extensively documented (triple-balanced recruitment with detailed methodology). anomalies field absent but quality control documented in preprocessing.

Fields:

Rationale: version field absent despite updates documenting specific versions (pilot, v1.0.0, v2.0.0, v3.0.0).

Fields:

Rationale: version_access field absent. No documentation of how to access previous versions or version-specific downloads.

Fields:

Rationale: updates field comprehensive with release timeline (pilot May 2024, v1.0.0 Nov 2024, ongoing through Nov 2026). errata field absent.

Fields:

Rationale: Update frequency documented: periodic releases during enrollment (through Nov 2026), final release late 2026. Specific release dates provided for pilot and v1.0.0.

Fields:

Rationale: was_derived_from and release_notes fields absent but provenance documented through collection_mechanisms, acquisition_methods, and preprocessing_strategies.

Fields:

Rationale: purposes comprehensively documented with 3 items: T2DM salutogenesis understanding, AI/ML standards establishment, addressing demographic inequities. Each with detailed descriptions.

Fields:

Rationale: tasks well-documented with 3 items: multi-domain AI/ML analyses, unbiased model development, disease trajectory studies. Clear alignment with purposes.

Fields:

Rationale: funders documented with NIH Common Fund Bridge2AI Program, primary grant OT2OD032644, additional grants (P30DK035816, UL1TR003096), Research to Prevent Blindness.

Fields:

Rationale: Grant numbers present: OT2OD032644 (primary), P30DK035816, UL1TR003096. Funding amount ($5,026,499 in 2022), opportunity number (OTA-21-008), and project dates (Sept 1, 2022 - Aug 31, 2025) documented.

Fields:

Rationale: creators extensively documented with 20 investigators including Contact PI/Project Leader (Aaron Lee), PIs, Co-Investigators with institutional affiliations and roles. Funders also documented.

Fields:

Rationale: collection_mechanisms comprehensively documented with 5 mechanisms: in-person visits (2.5-4 hours, 3 sites), EHR screening (ICD-10 codes), wave-based recruitment, home wearable monitoring (10 days), biospecimen collection/banking.

Fields:

Rationale: acquisition_methods extensively documented with 12 methods covering all data domains: surveys (REDCap), physical measurements, retinal imaging (7 devices), visual function, clinical labs (local and central), ECG, cognitive testing (MoCA), neuropathy (monofilament), continuous glucose (Dexcom G6), activity (Garmin), environmental sensors, biospecimens.

Fields:

Rationale: preprocessing_strategies (5 items) and cleaning_strategies (1 item) comprehensively documented. labeling_strategies absent but not applicable for this observational dataset.

Fields:

Rationale: software_and_tools field absent despite extensive software/equipment use documented in acquisition_methods (REDCap, DICOM tools, various imaging devices, Dexcom, Garmin, MoCA Duo).

Fields:

Rationale: external_resources extensively documented (9 resources). conforms_to absent but standards mentioned in preprocessing (OMOP CDM, DICOM, mHealth, RxNorm, ICD-10).

Fields:

Rationale: known_limitations field absent. Limitations implied in discouraged_uses (ongoing enrollment, need for clinical validation) but not formally documented.

Fields:

Rationale: known_biases field absent. No documentation of potential biases despite targeted recruitment and exclusion criteria.

Fields:

Rationale: anomalies field absent. Quality control documented but no specific anomalies, data quality issues, or missing data patterns described.

Fields:

Rationale: sensitive_elements comprehensively documented with 3 items: genetic/biospecimen data, geographic/demographic identifiers, medical history/records. content_warnings absent.

Fields:

Rationale: ethical_reviews absent as structured field but extensively covered in human_subject_research (IRB approvals, community advisory board, bioethics guidance). No conflicts of interest documented.

Fields:

Rationale: publisher field absent. Dataset hosted on FAIRhub and Zenodo but no formal publisher designation.

Fields:

Rationale: Both citation and doi fields absent. DOI exists (10.5281/zenodo.10642459) but only in external_resources, not as top-level field.

Fields:

Rationale: conforms_to field absent despite extensive standards use (OMOP CDM, DICOM, mHealth, FAIR principles).

Fields:

Rationale: external_resources extensively documented (9 resources) including project website, documentation, publications, data sharing policies. page field present.

Fields:

Rationale: related_datasets field absent. No typed relationships to other Bridge2AI datasets, prior T2DM studies, or derived datasets.