Dataset Information

Genome-Wide Association Studies-Based Machine Learning for Prediction of Age-Related Macular Degeneration Risk.

ABSTRACT:

Purpose

Because age-related macular degeneration (AMD) is a progressive disorder and advanced AMD is currently hard to cure, an accurate and informative prediction of a person's AMD risk using genetic information is desirable for early diagnosis and potential individualized clinical management. The objective of this study was to develop and validate novel prediction models for AMD risk using large genome-wide association studies datasets with different machine learning approaches.

Methods

Genotype data from 32,215 Caucasian individuals with age of ≥50 years from the International AMD Genomics Consortium in dbGaP were used to establish and test prediction models for AMD risk. Four different machine learning approaches-neural network, lasso regression, support vector machine, and random forest-were implemented. A standard logistic regression model using a genetic risk score was also considered.

Results

All machine learning-based methods achieved satisfactory performance for predicting advanced AMD cases (vs. normal controls) (area under the curve = 0.81-0.82, Brier score = 0.17-0.18 in a separate test dataset) and any stage AMD (vs. normal controls) (area under the curve = 0.78-0.79, Brier score = 0.18-0.20 in a separate test dataset). The prediction performance was further validated in an independent dataset of 783 subjects from UK Biobank (area under the curve = 0.67).

Conclusions

By applying multiple state-of-art machine learning approaches on large AMD genome-wide association studies datasets, the predictive models we established can provide an accurate estimation of an individual's AMD risk profile based on genetic information along with age. The online prediction interface is available at: https://yanq.shinyapps.io/no_vs_amd_NN/.

Translational relevance

The accurate and individualized risk prediction model interface will greatly improve early diagnosis and enhance tailored clinical management of AMD.

SUBMITTER: Yan Q

PROVIDER: S-EPMC7900884 | biostudies-literature |

REPOSITORIES: biostudies-literature

ACCESS DATA

Similar Datasets

Project description:PurposeTo determine if plasma metabolic profiles can detect differences between patients with neovascular age-related macular degeneration (NVAMD) and similarly-aged controls.MethodsMetabolomic analysis using liquid chromatography with Fourier-transform mass spectrometry (LC-FTMS) was performed on plasma samples from 26 NVAMD patients and 19 controls. Data were collected from mass/charge ratio (m/z) 85 to 850 on a Thermo LTQ-FT mass spectrometer, and metabolic features were extracted using an adaptive processing software package. Both non-transformed and log2 transformed data were corrected using Benjamini and Hochberg False Discovery Rate (FDR) to account for multiple testing. Orthogonal Partial Least Squares-Discriminant Analysis was performed to determine metabolic features that distinguished NVAMD patients from controls. Individual m/z features were matched to the Kyoto Encyclopedia of Genes and Genomes database and the Metlin metabolomics database, and metabolic pathways associated with NVAMD were identified using MetScape.ResultsOf the 1680 total m/z features detected by LC-FTMS, 94 unique m/z features were significantly different between NVAMD patients and controls using FDR (q = 0.05). A comparison of these features to those found with log2 transformed data (n = 132, q = 0.2) revealed 40 features in common, reaffirming the involvement of certain metabolites. Such metabolites included di- and tripeptides, covalently modified amino acids, bile acids, and vitamin D-related metabolites. Correlation analysis revealed associations among certain significant features, and pathway analysis demonstrated broader changes in tyrosine metabolism, sulfur amino acid metabolism, and amino acids related to urea metabolism.ConclusionsThese data suggest that metabolomic analysis can identify a panel of individual metabolites that differ between NVAMD cases and controls. Pathway analysis can assess the involvement of certain metabolic pathways, such as tyrosine and urea metabolism, and can provide further insight into the pathophysiology of AMD.

Project description:PurposeNeovascular age-related macular degeneration (nAMD) shows variable treatment response to intravitreal anti-VEGF. This analysis compared the potential of different artificial intelligence (AI)-based machine learning models using OCT and clinical variables to accurately predict at baseline the best-corrected visual acuity (BCVA) at 9 months in response to ranibizumab in patients with nAMD.DesignRetrospective analysis.ParticipantsBaseline and imaging data from patients with subfoveal choroidal neovascularization secondary to age-related macular dengeration.MethodsBaseline data from 502 study eyes from the HARBOR (NCT00891735) prospective clinical trial (monthly ranibizumab 0.5 and 2.0 mg arms) were pooled; 432 baseline OCT volume scans were included in the analysis. Seven models, based on baseline quantitative OCT features (Least absolute shrinkage and selection operator [Lasso] OCT minimum [min], Lasso OCT 1 standard error [SE]); on quantitative OCT features and clinical variables at baseline (Lasso min, Lasso 1SE, CatBoost, RF [random forest]); or on baseline OCT images only (deep learning [DL] model), were systematically compared with a benchmark linear model of baseline age and BCVA. Quantitative OCT features were derived by a DL segmentation model on the volume images, including retinal layer volumes and thicknesses, and retinal fluid biomarkers, including statistics on fluid volume and distribution.Main outcome measuresPrognostic ability of the models was evaluated using coefficient of determination (R2) and median absolute error (MAE; letters).ResultsIn the first cross-validation split, mean R2 (MAE) of the Lasso min, Lasso 1SE, CatBoost, and RF models was 0.46 (7.87), 0.42 (8.43), 0.45 (7.75), and 0.43 (7.60), respectively. These models ranked higher than or similar to the benchmark model (mean R2, 0.41; mean MAE, 8.20 letters) and better than OCT-only models (mean R2: Lasso OCT min, 0.20; Lasso OCT 1SE, 0.16; DL, 0.34). The Lasso min model was selected for detailed analysis; mean R2 (MAE) of the Lasso min and benchmark models for 1000 repeated cross-validation splits were 0.46 (7.7) and 0.42 (8.0), respectively.ConclusionsMachine learning models based on AI-segmented OCT features and clinical variables at baseline may predict future response to ranibizumab treatment in patients with nAMD. However, further developments will be needed to realize the clinical utility of such AI-based tools.Financial disclosuresProprietary or commercial disclosure may be found after the references.

Project description:Genome-wide association studies provide important insights to the genetic component of disease risks. However, an existing challenge is how to incorporate collective effects of interactions beyond the level of independent single nucleotide polymorphism (SNP) tests. While methods considering each SNP pair separately have provided insights, a large portion of expected heritability may reside in higher-order interaction effects.We describe an inference approach (discrete discriminant analysis; DDA) designed to probe collective interactions while treating both genotypes and phenotypes as random variables. The genotype distributions in case and control groups are modeled separately based on empirical allele frequency and covariance data, whose differences yield disease risk parameters. We compared pairwise tests and collective inference methods, the latter based both on DDA and logistic regression. Analyses using simulated data demonstrated that significantly higher sensitivity and specificity can be achieved with collective inference in comparison to pairwise tests, and with DDA in comparison to logistic regression. Using age-related macular degeneration (AMD) data, we demonstrated two possible applications of DDA. In the first application, a genome-wide SNP set is reduced into a small number (?100) of variants via filtering and SNP pairs with significant interactions are identified. We found that interactions between SNPs with highest AMD association were epigenetically active in the liver, adipocytes, and mesenchymal stem cells. In the other application, multiple groups of SNPs were formed from the genome-wide data and their relative strengths of association were compared using cross-validation. This analysis allowed us to discover novel collections of loci for which interactions between SNPs play significant roles in their disease association. In particular, we considered pathway-based groups of SNPs containing up to ?10, 000 variants in each group. In addition to pathways related to complement activation, our collective inference pointed to pathway groups involved in phospholipid synthesis, oxidative stress, and apoptosis, consistent with the AMD pathogenesis mechanism where the dysfunction of retinal pigment epithelium cells plays central roles.The simultaneous inference of collective interaction effects within a set of SNPs has the potential to reveal novel aspects of disease association.