Related Experiment Video
Updated: Jun 21, 2026
14:27
Identification of Disease-related Spatial Covariance Patterns using Neuroimaging Data
Published on: June 26, 2013
16.0K
Predictive analysis methods for human microbiome data with application to Parkinson's disease.
Mei Dong1, Longhai Li2, Man Chen2
1Dalla Lana School of Public Health, University of Toronto, Toronto, ON, Canada.
Plos One
|August 25, 2020
Summary
Predictive modeling of the gut microbiome for Parkinson's disease shows high accuracy. Machine learning methods like LASSO effectively identify key operational taxonomic units (OTUs) associated with the disease, offering new diagnostic potential.
Area of Science:
- Microbiome research
- Computational biology
- Statistical modeling
Background:
- Microbiome data (OTU counts) exhibit zero-inflation and over-dispersion, complicating traditional statistical testing.
- Current methods rely on p-values/q-values, which are sensitive to model assumptions and don't guarantee predictive power.
- Predictive analysis offers an alternative for identifying associated OTUs and assessing phenotype predictability.
Purpose of the Study:
- To investigate and compare three predictive analysis strategies for microbiome data.
- To evaluate the performance of these strategies on simulated and real-world Parkinson's disease data.
- To assess the predictive capability of gut microbiome composition for Parkinson's disease.
Main Methods:
- Three strategies were compared: LASSO regression, screening+GLM, and screening+LASSO.
- LASSO involved fitting a multinomial logistic regression model to transformed OTU counts.
- Screening methods utilized q-values from generalized linear mixed models (GLMM) for OTU selection before fitting generalized linear models (GLM) or LASSO.
- Empirical studies used Dirichlet-multinomial simulated datasets and real gut microbiome data from Parkinson's disease patients.
Main Results:
- LASSO with appropriate variable transformation demonstrated strong predictive performance on zero-inflated microbiome data.
- The screening+LASSO strategy, using binary transformed OTUs, age, and sex, predicted Parkinson's disease with high accuracy (Error Rate = 0.199, AUC = 0.872, AUPRC = 0.912).
- Simulation studies confirmed the effectiveness of LASSO for zero-inflated data.
Conclusions:
- Predictive modeling, particularly LASSO, is a powerful approach for analyzing complex microbiome data.
- The gut microbiome composition, alongside age and sex, can accurately predict Parkinson's disease.
- These findings strongly support a significant relationship between the gut microbiome and Parkinson's disease.
Related Concept Videos
01:22Introduction to the Human Microbiota
Microorganisms colonize various regions of the human body, including the mouth, nasal passages, throat, stomach, intestines, urogenital tract, and skin. The total number of microbial cells is estimated to range from 10¹³ to 10¹⁴—comparable to, or exceeding, the number of human somatic cells. This host–microbiome relationship has led to the conceptualization of humans as supraorganisms, wherein microbial communities perform vital roles in development, immunity, and disease...
01:24Parkinson Disease ll: Pathophysiology
Parkinson disease (PD) is a progressive neurodegenerative disorder primarily affecting movement, with additional non-motor features. Its pathophysiology involves complex interactions among genetic susceptibility, environmental exposures, and cellular dysfunction, including dopaminergic neuron loss, protein aggregation, and mitochondrial impairment.Selective NeurodegenerationA key feature is the degeneration of dopaminergic neurons in the substantia nigra pars compacta, leading to reduced...