Automated machine learning for endemic active tuberculosis prediction from multiplex serological data
Problem
Active tuberculosis serological diagnosis requires interpreting complex multi-antibody data across variable immune responses, and building suitable algorithms manually is slow.
Why it matters
Accurate TB diagnosis in endemic settings determines timely treatment of a major killer.
What the AI did
The automated machine learning platform MILO integrated data processing, feature selection, model training and validation to generate and evaluate thousands of models, tested on out-of-sample secondary and tertiary datasets.
Results
23-antigen model: 90.5% sensitivity; specificity 100.0% (TB vs healthy) and 74.6% (TB vs COPD)
Impact
Peer-reviewed model validated on two independent datasets; clinical deployment not stated.
Record
| Venue | Scientific Reports |
|---|---|
| Published | 2021 |
| Work type | article |
| Authors | Hooman H. Rashidi, Luke T. Dang, Samer Albahra, Resmi K. Ravindran, Imran H. Khan |
| Citations | 22 (OpenAlex, at import time) |
| References | 25 |
| Model or method | MILO (Machine Intelligence Learning Optimizer) |
| Organization | not stated in source |
| Primary source | https://doi.org/10.1038/s41598-021-97453-7 |
| Open access | gold |
| Keywords | tuberculosis, automated machine learning, serology, MILO, antigen model |
Importance rating
significant — thousands to millions affected
TB serodiagnosis in variable immune responses; abstract frames rapid clinical implementation as critical but states no disease burden scale.