# Why ML-based cough models do not generalize: a systematic cross-dataset evaluation for tuberculosis screening

arXiv: https://arxiv.org/abs/2608.25846 · HTML: https://mlsanity.com/paper/2608.25846/

## Key facts

- No citing papers in the map yet (data through 14 Oct 2026).
- Builds on 22 of its 48 references that are in the map.
- First posted on arXiv 26 Aug 2026.

## Builds on (most cited first, 22 of 22)

- [Deep Residual Learning for Image Recognition](https://mlsanity.com/paper/1512.03385/) (2015)
- [Explaining and Harnessing Adversarial Examples](https://mlsanity.com/paper/1412.6572/) (2014)
- [mixup: Beyond Empirical Risk Minimization](https://mlsanity.com/paper/1710.09412/) (2017)
- [Domain-Adversarial Training of Neural Networks](https://mlsanity.com/paper/1505.07818/) (2015)
- [Generalizing to Unseen Domains: A Survey on Domain Generalization](https://mlsanity.com/paper/2103.03097/) (2021)
- [CNN Architectures for Large-Scale Audio Classification](https://mlsanity.com/paper/1609.09430/) (2016)
- [Domain Generalization via Invariant Feature Representation](https://mlsanity.com/paper/1301.2115/) (2013)
- [Deep Domain Confusion: Maximizing for Domain Invariance](https://mlsanity.com/paper/1412.3474/) (2014)
- [Generalizing Across Domains via Cross-Gradient Training](https://mlsanity.com/paper/1804.10745/) (2018)
- [Scatter Component Analysis: A Unified Framework for Domain Adaptation and Domain Generalization](https://mlsanity.com/paper/1510.04373/) (2015)
- [MixStyle Neural Networks for Domain Generalization and Adaptation](https://mlsanity.com/paper/2107.02053/) (2021)
- [The COUGHVID crowdsourcing dataset: A corpus for the study of large-scale cough analysis algorithms](https://mlsanity.com/paper/2009.11644/) (2020)
- [Detecting Shortcut Learning for Fair Medical AI using Shortcut Testing](https://mlsanity.com/paper/2207.10384/) (2022)
- [Towards Open Respiratory Acoustic Foundation Models: Pretraining and Benchmarking](https://mlsanity.com/paper/2406.16148/) (2024)
- [Automatic Cough Classification for Tuberculosis Screening in a Real-World Environment](https://mlsanity.com/paper/2103.13300/) (2021)
- [HeAR -- Health Acoustic Representations](https://mlsanity.com/paper/2403.02522/) (2024)
- [A Semi-Supervised Algorithm for Improving the Consistency of Crowdsourced Datasets: The COVID-19 Case Study on Respiratory Disorder Classification](https://mlsanity.com/paper/2209.04360/) (2022)
- [Transformation of audio embeddings into interpretable, concept-based representations](https://mlsanity.com/paper/2504.14076/) (2025)
- [Optimizing Audio Augmentations for Contrastive Learning of Health-Related Acoustic Signals](https://mlsanity.com/paper/2309.05843/) (2023)
- [DG-SED: Domain Generalization for Sound Event Detection with Heterogeneous Training Data](https://mlsanity.com/paper/2407.03654/) (2024)
- [Acoustical Features as Knee Health Biomarkers: A Critical Analysis](https://mlsanity.com/paper/2405.15085/) (2024)
- [Tuberculosis Screening from Cough Audio: Baseline Models, Clinical Variables, and Uncertainty Quantification](https://mlsanity.com/paper/2601.07969/) (2026)

## Abstract (arXiv)

Cough acoustics are promising for non-invasive tuberculosis (TB) screening, yet whether machine learning (ML) models capture disease-related acoustics or artifacts of data collection remains unresolved. We evaluated the cross-dataset generalizability of classical ML and deep learning (DL) cough-based TB classifiers across three independent datasets. Despite moderate within-dataset performance (ROC-AUC up to 0.755 ± 0.056), both pipelines fail to generalize, with external performance frequently below 0.6, indicating a possible limitation of the data. We further observed audio representations are organized by recording device and dataset rather than TB status, predicted TB probability tracks country-level prevalence in CODA, and device mismatch degrades transfer while device-diverse training improves it. Additionally, a clinical-variable baseline generalizes more consistently (ROC-AUC 0.655 - 0.711), indicating acquisition-specific variability is a stronger driver of poor generalizability than population shift. High within-dataset performance is not enough. External validation is essential before cough-based TB models are clinically ready.