name: 'hemoglobinopathy-analysis-agent' description: 'AI-powered analysis of hemoglobin disorders including sickle cell disease, thalassemias, and variant hemoglobins using HPLC, electrophoresis, and molecular data.' measurable_outcome: Execute skill workflow successfully with valid output within 15 minutes. allowed-tools:

• read_file
• run_shell_command

Hemoglobinopathy Analysis Agent

The Hemoglobinopathy Analysis Agent provides comprehensive AI-driven analysis of hemoglobin disorders. It integrates HPLC chromatograms, electrophoresis patterns, CBC parameters, and molecular genetics for diagnosis and management of sickle cell disease, thalassemias, and variant hemoglobins.

When to Use This Skill

• When interpreting HPLC hemoglobin chromatograms for variant identification.
• To diagnose and classify thalassemia syndromes (α, β, δβ).
• For comprehensive sickle cell disease phenotype assessment.
• When correlating genotype with clinical phenotype severity.
• To guide hydroxyurea dosing and transfusion management.

Core Capabilities

1. HPLC Interpretation: AI pattern recognition for hemoglobin variant identification from HPLC chromatograms.

2. Thalassemia Classification: Distinguish α-thalassemia (silent carrier to Hb Bart's) and β-thalassemia (minor to major).

3. Sickle Cell Phenotyping: Integrate HbS%, HbF%, α-globin status for phenotype prediction.

4. Variant Identification: Database matching for >1,500 known hemoglobin variants.

5. Molecular Correlation: Link genetic variants (HBB, HBA1/2) to protein phenotypes.

6. Management Guidance: Treatment recommendations based on disease severity.

Hemoglobin Pattern Analysis

Workflow

1. Input: HPLC chromatogram, CBC with indices, peripheral smear findings, molecular data (if available).

2. Pattern Recognition: AI analysis of HPLC retention times and peak areas.

3. Variant Matching: Compare against hemoglobin variant database.

4. RBC Correlation: Integrate MCV, MCH, RDW, reticulocyte count.

5. Phenotype Classification: Assign clinical phenotype category.

6. Management: Generate treatment and monitoring recommendations.

7. Output: Diagnosis, variant identification, clinical classification, management plan.

Example Usage

User: "Interpret this HPLC chromatogram showing an abnormal peak and correlate with the CBC findings."

Agent Action:

python3 Skills/Hematology/Hemoglobinopathy_Analysis_Agent/hb_analyzer.py \
    --hplc_data chromatogram.csv \
    --retention_times peak_times.json \
    --cbc cbc_results.json \
    --peripheral_smear smear_findings.txt \
    --molecular hbb_sequencing.vcf \
    --output hb_report.json

Key Hemoglobin Variants

AI/ML Components

HPLC Pattern Recognition:

• CNN trained on 50,000+ chromatograms
• Identifies peaks by retention time and shape
• Quantifies hemoglobin fractions
• Flags unusual patterns for review

Phenotype Prediction:

• Gradient boosting model
• Features: Hb%, HbF%, α-globin genotype, F-cell distribution
• Predicts clinical severity (mild/moderate/severe)
• VOC risk, stroke risk, TCD velocity correlation

Genotype-Phenotype Correlation:

• Database of published correlations
• Modifier genes (BCL11A, HBS1L-MYB, α-globin)
• Pharmacogenomics (HU response prediction)

Clinical Decision Support

Hydroxyurea Candidacy:

• Severe phenotype
• ≥3 pain crises/year
• ACS history
• Stroke prevention

Transfusion Protocols:

• Simple vs exchange transfusion
• Target HbS% thresholds
• Iron chelation monitoring

Monitoring Schedule:

• LDH, reticulocytes, bilirubin
• Ferritin for transfused patients
• TCD for children with SCD

Prerequisites

• Python 3.10+
• PyTorch for image/signal analysis
• Hemoglobin variant databases
• Clinical lab interface

Related Skills

• Blood_Smear_Analysis - For morphology assessment
• Variant_Interpretation - For molecular findings
• Flow_Cytometry_AI - For F-cell quantification

Newborn Screening Integration

• Interpret newborn screening HPLC patterns
• Distinguish FAS (sickle trait) from FS (sickle disease)
• Flag FAE (HbE), FAC (HbC), F-only (β-thal major)
• Generate confirmatory testing recommendations

Author

AI Group - Biomedical AI Platform