mAML: an automated machine learning pipeline with a microbiome repository for human disease classification
Problem
Building machine learning models from human microbiome data for disease classification is tedious and requires manual ML expertise.
Why it matters
Microbiome-based prediction could enable personalized diagnosis of human diseases.
What the AI did
mAML, an AutoML pipeline on a web platform, automatically generates optimized, interpretable microbiome-based classification models reproducibly.
Results
High performance on 13 benchmark datasets (binary and multi-class); no numeric metrics stated in abstract.
Impact
Provides a scalable web tool plus the GMrepo ML repository for microbiome disease research.
Record
| Venue | Database |
|---|---|
| Published | 2020 |
| Work type | article |
| Authors | Fenglong Yang, Quan Zou |
| Citations | 38 (OpenAlex, at import time) |
| References | 38 |
| Model or method | mAML (AutoML pipeline) |
| Organization | not stated in source |
| Primary source | https://doi.org/10.1093/database/baaa050 |
| Open access | gold |
| Keywords | AutoML, microbiome, disease classification, benchmark datasets, web platform, GMrepo |
Importance rating
real but narrower advance
Abstract states the repository covers 120 classification tasks, 85 human-disease phenotypes, 12,429 metagenomic and 38,643 amplicon samples; a useful but narrower tooling advance for microbiome research.