How a Student‑Led AI System Accelerated COVID‑19 Diagnosis in Seconds

At the onset of the 2020 COVID‑19 outbreak, a group of Huazhong University of Science and Technology students and faculty formed an AI war team to address the shortage of rapid diagnostic tools.

Within 11 days, the team developed an AI quantitative medical assistance system that could analyze lung CT images in seconds, helping dozens of Chinese hospitals diagnose COVID‑19 patients quickly.

The project received strong endorsement from Vice‑Premier Sun Chunlan and was later promoted to dozens of countries, eventually reaching the finals of the 6th China International "Internet+" University Innovation Competition.

Key challenges included gathering massive CT datasets (300‑500 images per patient) under lockdown conditions, which the team overcame by physically transporting data from hospitals and using Huawei's WeLink for real‑time collaboration.

Data were managed on Huawei Cloud Object Storage Service (OBS), enabling efficient labeling of over 4,800 cases by a distributed team of up to 68 members.

Huawei also provided the ModelArts platform for model training and testing, allowing the team to monitor loss and segmentation results via a visual interface, dramatically reducing development time.

The system offers three core functions: (1) instant COVID‑19 detection from CT scans, (2) quantitative severity assessment to distinguish mild from severe cases, and (3) prediction of mild cases progressing to severe disease using clinical data and imaging.

Compared with manual reading (15‑30 minutes per case), the AI system delivers results in 2‑3 seconds, providing fast and precise information for clinicians.

To adapt to evolving pandemic needs, the team later incorporated federated learning on Huawei Cloud, enabling multi‑institution model training while preserving data privacy.

Looking ahead, the team plans to extend the AI platform to other diseases such as pancreatic cancer, lung nodules, and X‑ray imaging, ensuring the system remains valuable beyond the COVID‑19 crisis.