Fabry Disease: AI-Extracted Clinical Dataset from Published Case Reports

General Information Title: Fabry Disease: AI-Extracted Clinical Dataset from Published Case Reports Version: 1.0.0 Date of export: 2026-02-27 Created by: Silene Systems (AI-powered clinical data extraction pipeline) Disease: Fabry Disease (OMIM #301500) Gene: GLA (Xq22.1) Inheritance: X-linked Description This dataset contains structured clinical data for 154 patients with Fabry disease, automatically extracted from 93 peer-reviewed case reports and case series indexed in PubMed Central. Data was extracted using a multi-agent AI pipeline with automated validation, clinical plausibility checks, and human review. Each patient record includes demographics, genetic variants, clinical symptoms with temporal information, laboratory values, and links to source publications. Symptoms are provided in two formats: original AI-extracted terms and HPO (Human Phenotype Ontology)-normalized mappings with confidence scores. The dataset is intended for rare disease research, phenotype-genotype correlation studies, natural history analysis, and machine learning applications in clinical genetics. Dataset Summary Metric Value Patients 154 Publications 93 Symptom observations (original) 2,039 Symptom observations (HPO-normalized) 2,017 Unique symptom names 803 Lab value measurements 1,264 Patients with genetic variant 103 Unique genetic variants 60 Demographics Count Male 73 Female 73 Unknown sex 8 Age metric Available Min Max Median Age at symptom onset 54 0 74 36 Age at diagnosis 124 4 75 48 Phenotype Distribution Phenotype Count Classic 66 Later-onset 67 Asymptomatic carrier 1 File Descriptions patients.csv (154 rows) One row per patient. Contains demographics, genetics, and clinical summary. Column Type Description case_id string Unique patient identifier (e.g., FAB-001) sex string Patient sex (male/female) age_at_symptom_onset numeric Age in years at first symptom onset age_at_diagnosis numeric Age in years at Fabry disease diagnosis diagnostic_delay_years numeric Difference between diagnosis and onset age genetic_variant string GLA gene variant in HGVS notation (e.g., c.644A>G) protein_change string Protein-level change (e.g., p.Asn215Ser) zygosity string Hemizygous, heterozygous, or homozygous phenotype string Clinical phenotype: classic, later-onset, or asymptomatic carrier disease_stage string Disease stage if reported diagnosis_summary string AI-generated clinical summary extraction_confidence numeric AI extraction confidence score (0-1) data_completeness numeric Proportion of fields populated (0-1) publication_pmcid string PubMed Central ID of source publication symptoms.csv (2,039 rows) One row per symptom observation, sourced from the original AI extraction. Preserves the exact clinical terminology used in source publications. Column Type Description case_id string Patient identifier symptom_name string Original symptom name as extracted from the publication present boolean Whether the symptom is present (True) or explicitly absent (False) details string Additional clinical details or context timepoint_raw string Raw temporal reference from the publication timepoint_category string Normalized category: onset, diagnosis, follow_up, etc. age_at_event numeric Patient age at the time of this observation offset_months numeric Months relative to a reference event symptoms_hpo.csv (2,017 rows) One row per symptom mapped to the Human Phenotype Ontology (HPO). Enables standardized phenotype analysis and cross-dataset comparisons. HPO mappings have been validated and corrected using LLM-based quality review. Column Type Description case_id string Patient identifier original_name string Original symptom name from the publication canonical_name string HPO-normalized term name hpo_id string HPO identifier (e.g., HP:0001639) present boolean Whether the symptom is present (True) or absent (False) details string Additional clinical details timepoint string Temporal reference context string Clinical context confidence numeric Mapping confidence score (0-1) lab_values.csv (1,264 rows) One row per laboratory measurement or diagnostic test result. Column Type Description case_id string Patient identifier lab_name string Name of the laboratory test or biomarker value string Measured value (numeric or descriptive) unit string Unit of measurement context string Clinical context (e.g., baseline, post-treatment) timepoint_raw string Raw temporal reference timepoint_category string Normalized temporal category age_at_event numeric Patient age at measurement offset_months numeric Months relative to a reference event publications.csv (93 rows; 95 with header) One row per source publication. Column Type Description pmcid string PubMed Central ID pmid string PubMed ID doi string Digital Object Identifier title string Publication title authors string Author list (semicolon-separated) journal string Journal name publication_date date Publication date (YYYY-MM-DD) license string Publication license dataset_summary.json Machine-readable summary with aggregate statistics including patient counts, demographic distributions, genetic variant counts, and phenotype breakdown. Methodology Data Extraction Pipeline Publication Discovery — Automated PubMed search for Fabry disease case reports (publications from 2015 onwards) Relevance Screening — AI-based abstract screening to filter relevant case reports Case Extraction — Structured patient data extraction from full-text articles using Claude Opus 4.6 (Claude, Anthropic) Clinical Validation — Automated checks for temporal logic, lab value plausibility, and genetic consistency against disease-specific profiles HPO Normalization — Mapping of extracted symptom terms to Human Phenotype Ontology, with LLM-based validation and correction of mappings Timepoint Normalization — Structuring of free-text temporal references into standardized categories with chronological ordering Quality Assurance Extraction confidence and data completeness scores per patient Automated clinical plausibility validation HPO mapping quality validated by LLM review Negated findings (symptoms explicitly absent) are captured with present=False Manual audit of random samples against source publications confirmed >90% symptom completeness and near-complete laboratory value extraction Limitations Data quality depends on the detail and clarity of source publications Some clinical findings may be missed if not explicitly stated in the text HPO mappings for ~246 symptoms lack HPO IDs (canonical name provided but ID not in reference ontology) Temporal information is not always available or precisely reported in source publications Dataset is limited to publications indexed in PubMed Central with available full text License The structured data in this dataset is derived from published scientific literature. Individual source publications retain their original licenses (see publications.csv). This compiled dataset is made available under CC BY 4.0.