Document: D4D - AI-READI FAIRHub.docx
==================================================

PARAGRAPHS:
--------------------------------------------------
[0] Style: normal
Text: # This YAML follows the "data-sheets-schema" LinkML structure

[2] Style: normal
Text: id: "https://doi.org/10.60775/fairhub.2"

[3] Style: normal
Text: title: "Flagship Dataset of Type 2 Diabetes from the AI-READI Project"

[4] Style: normal
Text: description: >

[5] Style: normal
Text: This dataset contains data from 1067 participants (collected between July 19,

[6] Style: normal
Text: 2023 and July 31, 2024) as part of the AI-READI project. Data are multimodal,

[7] Style: normal
Text: contain no PHI, and specifically exclude sex, race/ethnicity, and medication

[8] Style: normal
Text: information to enhance privacy protection. Its size is approximately 2.01 TB

[9] Style: normal
Text: across 165,051 files. The dataset is intended to enable AI/ML research on

[10] Style: normal
Text: type 2 diabetes mellitus (T2DM), focusing on disease trajectories, salutogenic

[11] Style: normal
Text: pathways, and equitable research practices. See docs.aireadi.org (v2.0.0) for

[12] Style: normal
Text: detailed descriptions, data structure, and usage guidelines.

[14] Style: normal
Text: license: >

[15] Style: normal
Text: Custom "Health Data License" (see https://doi.org/10.5281/zenodo.10642459)

[16] Style: normal
Text: version: "2.0.0"

[17] Style: normal
Text: doi: "doi:10.60775/fairhub.2"

[18] Style: normal
Text: keywords:

[19] Style: normal
Text: - "Diabetes mellitus"

[20] Style: normal
Text: - "Machine Learning"

[21] Style: normal
Text: - "Artificial Intelligence"

[22] Style: normal
Text: - "Electrocardiography"

[23] Style: normal
Text: - "Continuous Glucose Monitoring"

[24] Style: normal
Text: - "Retinal imaging"

[25] Style: normal
Text: - "Eye exam"

[26] Style: normal
Text: bytes: 2010000000000  # Approximate byte size for 2.01 TB

[27] Style: normal
Text: is_tabular: false     # Dataset is multimodal, not strictly tabular

[29] Style: normal
Text: ## MOTIVATION-RELATED FIELDS

[30] Style: normal
Text: purposes:

[31] Style: normal
Text: - response: >

[32] Style: normal
Text: The dataset was created to support AI/ML research into type 2 diabetes

[33] Style: normal
Text: (T2DM), with a focus on understanding complex disease trajectories and

[34] Style: normal
Text: salutogenic pathways.

[35] Style: normal
Text: tasks:

[36] Style: normal
Text: - response: >

[37] Style: normal
Text: Primary tasks include disease staging, progression analysis, manifold

[38] Style: normal
Text: modeling for pseudo-time analysis, and other T2DM-related analytics.

[39] Style: normal
Text: addressing_gaps:

[40] Style: normal
Text: - response: >

[41] Style: normal
Text: It addresses a lack of large, high-quality, ethically sourced, and

[42] Style: normal
Text: inclusively designed multimodal datasets in the T2DM domain.

[43] Style: normal
Text: creators:

[44] Style: normal
Text: - principal_investigator:

[45] Style: normal
Text: name: "AI-READI Consortium"

[46] Style: normal
Text: affiliation: "Multiple research institutions funded by NIH Bridge2AI Common Fund"

[47] Style: normal
Text: funders:

[48] Style: normal
Text: - grantor:

[49] Style: normal
Text: id: "NIH-Bridge2AI-CommonFund"

[50] Style: normal
Text: name: "NIH Bridge2AI Common Fund"

[51] Style: normal
Text: grant:

[52] Style: normal
Text: name: "Bridge2AI - AI-READI"

[53] Style: normal
Text: grant_number: "1OT2OD032644"

[55] Style: normal
Text: ## COMPOSITION-RELATED FIELDS

[56] Style: normal
Text: instances:

[57] Style: normal
Text: - data_topic: "B2AI_TOPIC:xxx"         # Example placeholder

[58] Style: normal
Text: instance_type: "Participant-based multimodal data"

[59] Style: normal
Text: data_substrate: "B2AI_SUBSTRATE:xxx" # Example placeholder

[60] Style: normal
Text: counts: 1067

[61] Style: normal
Text: label: false

[62] Style: normal
Text: sampling_strategies: []

[63] Style: normal
Text: missing_information: []

[64] Style: normal
Text: anomalies: []

[65] Style: normal
Text: confidential_elements:

[66] Style: normal
Text: - confidential_elements_present: false

[67] Style: normal
Text: description: []

[68] Style: normal
Text: content_warnings:

[69] Style: normal
Text: - content_warnings_present: false

[70] Style: normal
Text: warnings: []

[71] Style: normal
Text: subpopulations:

[72] Style: normal
Text: - subpopulation_elements_present: false

[73] Style: normal
Text: identification: []

[74] Style: normal
Text: distribution: []

[75] Style: normal
Text: sensitive_elements:

[76] Style: normal
Text: - sensitive_elements_present: true

[77] Style: normal
Text: description:

[78] Style: normal
Text: - >

[79] Style: normal
Text: Contains health data related to T2DM. However, all PHI and direct

[80] Style: normal
Text: identifiers (including race/ethnicity, sex, and medication usage) have

[81] Style: normal
Text: been removed to reduce sensitivity risks.

[82] Style: normal
Text: is_deidentified:

[83] Style: normal
Text: identifiable_elements_present: false

[84] Style: normal
Text: description:

[85] Style: normal
Text: - >

[86] Style: normal
Text: Directly identifying information (PHI) was removed. The study uses

[87] Style: normal
Text: participant ID numbers instead of names or other identifying details.

[89] Style: normal
Text: ## COLLECTION-RELATED FIELDS

[90] Style: normal
Text: acquisition_methods:

[91] Style: normal
Text: - description:

[92] Style: normal
Text: - >

[93] Style: normal
Text: Data gathered through direct clinical measures (e.g., ECG, wearable

[94] Style: normal
Text: devices), environmental sensors, and imaging techniques (OCT, FLIO,

[95] Style: normal
Text: etc.). The dataset includes cross-sectional data (N=1067) and a planned

[96] Style: normal
Text: longitudinal subset (~10% of participants).

[97] Style: normal
Text: was_directly_observed: true

[98] Style: normal
Text: was_reported_by_subjects: true

[99] Style: normal
Text: was_inferred_derived: false

[100] Style: normal
Text: was_validated_verified: true

[101] Style: normal
Text: collection_mechanisms:

[102] Style: normal
Text: - description:

[103] Style: normal
Text: - >

[104] Style: normal
Text: The data were collected via standardized clinical protocols, including

[105] Style: normal
Text: 12-lead ECG, wearable activity monitors, continuous glucose monitors,

[106] Style: normal
Text: retinal imaging (OCT, OCTA, FLIO, photography), and custom

[107] Style: normal
Text: environmental sensors. Clinical data were collected via REDCap forms.

[108] Style: normal
Text: data_collectors:

[109] Style: normal
Text: - description:

[110] Style: normal
Text: - >

[111] Style: normal
Text: Trained research coordinators and clinical staff from AI-READI project

[112] Style: normal
Text: sites collected data; participants were compensated. Some data points

[113] Style: normal
Text: were self-reported via questionnaires.

[114] Style: normal
Text: collection_timeframes:

[115] Style: normal
Text: - description:

[116] Style: normal
Text: - >

[117] Style: normal
Text: Data in this version (v2.0.0) covers collection from July 19, 2023 to

[118] Style: normal
Text: July 31, 2024. Additional data will be integrated in future releases.

[119] Style: normal
Text: ethical_reviews:

[120] Style: normal
Text: - description:

[121] Style: normal
Text: - >

[122] Style: normal
Text: The study was reviewed and approved by an Institutional Review Board

[123] Style: normal
Text: (IRB). For details, refer to the study_description.json and IRB

[124] Style: normal
Text: documentation at docs.aireadi.org.

[125] Style: normal
Text: data_protection_impacts:

[126] Style: normal
Text: - description:

[127] Style: normal
Text: - >

[128] Style: normal
Text: A data protection impact assessment was performed to ensure compliance

[129] Style: normal
Text: with relevant regulatory frameworks, focusing on secure handling and

[130] Style: normal
Text: anonymization of participant data.

[132] Style: normal
Text: ## PREPROCESSING-CLEANING-LABELING FIELDS

[133] Style: normal
Text: preprocessing_strategies:

[134] Style: normal
Text: - description:

[135] Style: normal
Text: - >

[136] Style: normal
Text: Data were harmonized to the Clinical Dataset Structure (CDS) v0.1.1

[137] Style: normal
Text: standard. Certain fields (e.g., race, sex) were removed for privacy.

[138] Style: normal
Text: cleaning_strategies:

[139] Style: normal
Text: - description:

[140] Style: normal
Text: - >

[141] Style: normal
Text: PHI was systematically removed. Missing values were left unaltered

[142] Style: normal
Text: unless irreparable, in which case records were flagged or removed to

[143] Style: normal
Text: preserve data integrity.

[144] Style: normal
Text: labeling_strategies:

[145] Style: normal
Text: - description:

[146] Style: normal
Text: - >

[147] Style: normal
Text: No explicit "label" fields were added; the dataset is not strictly

[148] Style: normal
Text: labeled for classification tasks, though certain columns in clinical

[149] Style: normal
Text: data map to relevant T2DM outcomes or statuses.

[151] Style: normal
Text: ## USES FIELDS

[152] Style: normal
Text: existing_uses:

[153] Style: normal
Text: - description:

[154] Style: normal
Text: - >

[155] Style: normal
Text: The dataset is the flagship dataset for AI-READI and is being used

[156] Style: normal
Text: internally for manifold modeling and T2DM subtyping analyses.

[157] Style: normal
Text: use_repository:

[158] Style: normal
Text: - description:

[159] Style: normal
Text: - >

[160] Style: normal
Text: Publications and software using this dataset are referenced in

[161] Style: normal
Text: https://zenodo.org/communities/aireadi and

[162] Style: normal
Text: https://github.com/AI-READI.

[163] Style: normal
Text: other_tasks:

[164] Style: normal
Text: - description:

[165] Style: normal
Text: - >

[166] Style: normal
Text: Possible uses include T2DM risk prediction, multi-organ disease

[167] Style: normal
Text: analysis, environmental correlations, and sensor-based research.

[168] Style: normal
Text: future_use_impacts:

[169] Style: normal
Text: - description:

[170] Style: normal
Text: - >

[171] Style: normal
Text: Dataset consumers should be mindful of potential biases in T2DM cohorts

[172] Style: normal
Text: and missing demographic details. Overgeneralizing results without

[173] Style: normal
Text: considering these aspects could lead to unfair or inaccurate outcomes.

[174] Style: normal
Text: discouraged_uses:

[175] Style: normal
Text: - description:

[176] Style: normal
Text: - >

[177] Style: normal
Text: Non-health-related, purely commercial exploitation or attempts to

[178] Style: normal
Text: reidentify participants are explicitly disallowed by the license terms.

[180] Style: normal
Text: ## DISTRIBUTION FIELDS

[181] Style: normal
Text: distribution_formats:

[182] Style: normal
Text: - description:

[183] Style: normal
Text: - >

[184] Style: normal
Text: Distributed via secure portal; data are organized by modality/device,

[185] Style: normal
Text: and each folder follows "Clinical Dataset Structure (CDS) v0.1.1."

[186] Style: normal
Text: distribution_dates:

[187] Style: normal
Text: - description:

[188] Style: normal
Text: - "v2.0.0 release date: 11/8/2024"

[189] Style: normal
Text: license_and_use_terms:

[190] Style: normal
Text: - description:

[191] Style: normal
Text: - >

[192] Style: normal
Text: Licensed under a custom Health Data License with obligations for data

[193] Style: normal
Text: usage, security, and restricted secondary sharing to protect

[194] Style: normal
Text: participants. See License file in the dataset or

[195] Style: normal
Text: https://doi.org/10.5281/zenodo.10642459

[196] Style: normal
Text: ip_restrictions:

[197] Style: normal
Text: - description:

[198] Style: normal
Text: - >

[199] Style: normal
Text: None beyond the custom license restrictions; no additional third-party

[200] Style: normal
Text: IP constraints reported.

[201] Style: normal
Text: regulatory_restrictions:

[202] Style: normal
Text: - description:

[203] Style: normal
Text: - >

[204] Style: normal
Text: Dataset usage is restricted to type 2 diabetes research only and must

[205] Style: normal
Text: comply with IRB terms and the Health Data License.

[207] Style: normal
Text: ## MAINTENANCE FIELDS

[208] Style: normal
Text: maintainers:

[209] Style: normal
Text: - description: ["Consortium", "Data coordinator staff"]

[210] Style: normal
Text: updates:

[211] Style: normal
Text: - description:

[212] Style: normal
Text: - >

[213] Style: normal
Text: New versions will integrate additional participants and potential data

[214] Style: normal
Text: corrections. Updates announced via docs.aireadi.org and consortium

[215] Style: normal
Text: mailing lists.

[216] Style: normal
Text: retention_limit:

[217] Style: normal
Text: description:

[218] Style: normal
Text: - >

[219] Style: normal
Text: Retention adheres to the IRB-approved protocol: data remain available for

[220] Style: normal
Text: the project duration and at least 5 years afterward; participants may

[221] Style: normal
Text: request removal.

[222] Style: normal
Text: version_access:

[223] Style: normal
Text: - description:

[224] Style: normal
Text: - >

[225] Style: normal
Text: v1.0.0 pilot data remains available but unsupported. v2.0.0 is current.

[226] Style: normal
Text: Future versions will maintain separate DOIs; older versions remain

[227] Style: normal
Text: archived.

[228] Style: normal
Text: extension_mechanism:

[229] Style: normal
Text: - description:

[230] Style: normal
Text: - >

[231] Style: normal
Text: External collaborators can propose changes via GitHub or contact forms.

[232] Style: normal
Text: Consortium reviews proposals for acceptance. Accepted contributions get

[233] Style: normal
Text: integrated and versioned.


TABLES:
--------------------------------------------------

DOCUMENT PROPERTIES:
--------------------------------------------------
Title: Word Document
Author: 
Created: None
Modified: 2025-04-16 17:44:58+00:00
