MiMIR

MiMIR projects pre-trained metabolomics-based models and imputes clinical and disease risk from 1H-NMR metabolomics data generated by Nightingale Health to enable interpretation of metabolic profiles in large epidemiological studies.

Key Features:

• Pre-trained Metabolomics-based Models: Includes 24 pre-trained models that can be projected and calibrated onto new datasets for estimation of disease risk and clinical risk factors.
• Model Projection and Calibration: Supports projection of existing risk models onto 1H-NMR datasets and calibration of model outputs to the target data.
• 1H-NMR Nightingale Health Data Support: Processes and analyzes 1H-NMR metabolomics data specifically from Nightingale Health assays.
• Statistical Analysis of Metabolic Data: Provides statistical analysis functionality for inspection and interpretation of metabolic profiles.
• Standardized Metabolic Profiling for Large Studies: Facilitates standardized, cost-effective metabolic profiling using 1H-NMR suitable for large-scale epidemiological studies.

Scientific Applications:

• Risk Imputation and Prediction: Imputes clinical and disease risk factors from 1H-NMR metabolic profiles using pre-trained models.
• Biomarker Exploration: Projects established risk models onto new cohorts to explore metabolic biomarkers associated with diseases and clinical endpoints.
• Population-scale Metabolic Profiling: Enables standardized metabolic profiling and cross-cohort comparisons in large epidemiological studies using Nightingale Health data.
• Model Transferability Assessment: Assesses and calibrates pre-trained metabolomics model performance when applied to novel datasets.

Methodology:

Projection of 24 pre-trained metabolomics-based models onto 1H-NMR datasets, calibration of model outputs, and statistical analysis of 1H-NMR metabolomics data from Nightingale Health for inspection and interpretation.