Adaptive federated learning with differential privacy for multi-class respiratory disease recognition from lung sound recordings
Syed Riyazul Haq, Kuruva Lakshmanna · 2026
Read Adaptive federated learning with differential privacy for multi-class respiratory disease recognition from lung sound recordings by Syed Riyazul Haq, Kuruva Lakshmanna (2026) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/pmc-13547319.txt.
Description
Haq and Lakshmanna build a federated system that keeps data private while it learns to spot lung diseases from breath sounds, in a study from twenty twenty-six. Training in one center would expose patient recordings and handle rare classes badly. Five fake hospital clients train as one, with no raw audio shared. Sound features are squeezed down, balanced, and sorted by a hybrid method.
Noise for privacy is added only when the models are merged, and in a clever spot. Accuracy and weighted scores reach about ninety-seven point six percent. Tests that drop one piece at a time show each piece beats simpler baselines. Rare classes gain from fake extra samples in the pipeline. Hospital networks and telemedicine are the settings the system aims at.
Catalog record
| CREATOR | Syed Riyazul Haq, Kuruva Lakshmanna |
|---|---|
| YEAR | 2026 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 44 KB |
| VERSION | 1.0 |
| LICENSE | CC BY 4.0 |
| RIGHTS | CC BY |
| SOURCE | Europe PMC |