AI and IoT Predict COPD Exacerbations Remotely

According to a JMIR Preprints preprint by Montenegro et al. (2026), the authors conducted a systematic review titled "AI and Internet of Things for Chronic Obstructive Pulmonary Disease Remote Monitoring: A Systematic Review of ECOPD Prediction Frameworks and Key Monitoring Physiological Variables". The preprint states the review maps both hardware (wearables, IoT sensors) and software (machine learning, deep learning) approaches used to predict exacerbations of COPD (ECOPD) and identifies the physiological and environmental variables commonly monitored (JMIR Preprints, 2026).

Per the preprint abstract, the review catalogs devices and sensor modalities that enable continuous remote monitoring, and it summarizes reported AI pipelines used for ECOPD prediction, including conventional machine learning and deep learning architectures (JMIR Preprints, 2026). The authors emphasize the challenge of processing heterogeneous physiological and environmental data streams captured by wearables and IoT platforms, which motivates the need for robust preprocessing and feature-extraction steps in the reviewed studies (JMIR Preprints, 2026).

In comparable remote-health research, heterogeneous sensor data, label scarcity for clinically verified exacerbations, and distribution shifts between study cohorts commonly complicate model development and external validation. Practitioners should note that combining temporal physiologic signals with contextual environmental data raises requirements for synchronization, missing-data handling, and lightweight on-device inference or reliable edge-cloud pipelines.

Systematic reviews like this one are useful for practitioners because they consolidate which sensors and data modalities attract the most evidence, and they summarize ML approaches that have been tried in the field. For teams designing COPD monitoring systems, the review can serve as a survey of candidate variables, device classes, and algorithmic patterns to evaluate against local clinical needs and deployment constraints.

Observers should look for the peer-reviewed version of the JMIR manuscript, studies providing prospective validation in real-world cohorts, emerging data standards for wearable-collected respiratory signals, and reproducible benchmarks that compare model performance across devices and populations.