An efficient mixture of deep and machine learning models for COVID-19 diagnosis in chest X-ray images
Problem
Screening and diagnosing COVID-19 patients quickly from chest X-rays to reduce the burden on medical experts.
Why it matters
Fast, accurate screening during a pandemic reduces healthcare workload and supports earlier treatment of infected patients.
What the AI did
Transfer learning with five pre-trained deep models (Xception best) was combined with machine-learning classification of deep features into an end-to-end diagnostic method.
Results
Xception reached 96.75% diagnostic accuracy; combined method tested on two datasets (further numbers truncated in abstract)
Impact
Evidence supports high reported accuracy on two datasets; clinical adoption not stated in source.
Record
| Venue | PLoS ONE |
|---|---|
| Published | 2020 |
| Work type | article |
| Authors | Dingding Wang, Jiaqing Mo, Gang Zhou, Liang Xu, Yajun Liu |
| Citations | 112 (OpenAlex, at import time) |
| References | 37 |
| Model or method | Xception plus combined deep features and machine learning classifiers |
| Organization | not stated in source |
| Primary source | https://doi.org/10.1371/journal.pone.0242535 |
| Open access | gold |
| Keywords | COVID-19, chest X-ray, Xception, transfer learning, computer-aided diagnosis |
Importance rating
major — millions of lives, an industry or a field
Abstract states COVID-19 seriously threatens human life and health worldwide; fast screening during a pandemic affects millions of lives and entire healthcare systems.