Objective: This study aims to reduce expert annotation effort in detecting patient-ventilator asynchrony (PVA) by introducing a semi-supervised learning framework for time series classification. Methods and procedures: We propose a model-independent framework that integrates hierarchical clustering and dynamic time-warping (DTW) for efficient data selection and label projection. The framework includes five steps: data collection, selection, annotation, projection, and model training. It is validated using a fully labeled dataset from Fondazione I.R.C.C.S. Policlinico San Matteo and applied to an unlabeled dataset from Maasstad Hospital, where annotation consistency and label quality are analyzed. Results: The framework reduces annotation effort by over 75% while closely resembling classification performance. On the San Matteo dataset, the model trained with projected labels achieved performance close to that of a fully supervised model. The method effectively captured rare PVA types and improved macro-averaged F1 scores compared to random sampling. On the Maasstad dataset, despite annotation inconsistencies, the framework demonstrated moderate detection performance (75% micro-averaged F₁ score) using labels from a single clinical expert. Conclusion: Our semi-supervised framework enables scalable and efficient annotation of clinical time series data, maintaining model accuracy with minimal expert input. It is robust across datasets and adaptable to varying signal quality and annotation consistency.

Mechanical ventilators are essential for patients who are unable to breathe independently. The aim of this article is to develop a systematic control design methodology that achieves accurate tracking of both the pressure and flow to ensure comfortable breathing for the patient. A hybrid controller is introduced that ensures improved baseflow tracking performance. The actual controller design leverages frequency-based techniques and is based on static decoupling and the factorized Nyquist criterion. Furthermore, a theoretical stability analysis of the hybrid controller is presented. The presented control strategy is implemented in a real ventilator, and it is demonstrated that the tracking performance is improved by conducting an experimental case study.

Mechanical ventilators are complex mechatronic devices that are essential for patients who are unable to breathe independently. The aim of this paper is to develop a systematic control method that achieves accurate tracking of both the pressure and flow to ensure comfortable breathing for the patient. This is achieved by using a feedback design procedure technique based on static decoupling and the factorized Nyquist criterion. Furthermore, switching controllers are introduced that allow for improved baseflow tracking performance. The presented control method is implemented in a real ventilator and it is demonstrated that the tracking performance is improved by conducting an experimental case-study.

Safe deployment of neural networks to classify time series in safety-critical applications relies on the ability of the classifier to detect data that does not originate from the same distribution as the training data. The aim of this paper is to propose a framework for detecting whether time series data is sampled from a different distribution than the training data, known as the problem of out-of-distribution (OOD) detection. We propose a novel distance-based OOD method for time series data using a hierarchical clustering method together with dynamic time-warping to measure the difference between a new data instance and the training set. The method is evaluated in the context of mechanical ventilation, a safety critical application, using both simulated and clinical datasets. Results of the mechanical ventilation use case demonstrate that the proposed approach effectively detects out-of-distribution data and improves classification performance in diverse settings.

Estimation of the breathing effort and relevant lung parameters of a ventilated patient is essential to keep track of a patient's clinical condition. The aim of this paper is to increase estimation accuracy through experiment design. The main method is an experiment design approach across multiple breaths within a linear regression framework to accurately identify the patient's condition. Identifiability and persistence of excitation are used to formulate an estimation problem with a unique solution. Furthermore, Fisher information is used for assessing the parameters sensitivity to slight changes of the ventilator settings to improve the variance of the estimation. The estimation method is applied to simulated patients who breathe regularly but also to patients who have variable breathing patterns. A virtual experiment is conducted for both situations to generate estimation results. The results are analyzed using mathematical tools and show that uniquely estimating the lung parameters and breathing effort over multiple breaths for both regularly and variably breathing patients is possible in the presented framework. The proposed estimation method obtains clinically relevant estimates for a large set of breathing disturbances from the simulation case-study.

Patient-ventilator asynchrony is one of the largest challenges in mechanical ventilation and is associated with prolonged ICU stay and increased mortality. The aim of this paper is to automatically detect and classify the different types of patient-ventilator asynchronies during a patient's breath using the typically available data on commercially available ventilators. This is achieved by a detection and classification framework using an objective definition of asynchrony and a supervised learning approach. The achieved detection performance of the near-real time framework on a clinical dataset is a significant improvement over current clinical practice, therewith and, this framework has the potential to significantly improve the patient comfort and treatment outcomes.

Estimation of relevant lung parameters and the breathing effort of a ventilated patient is essential to keep track of the patient's clinical condition. The aim of this paper is to investigate the major challenges of estimating the patient's condition with parametric models. The main method is a linear regression framework, where identifiability and persistence of excitation aspects are clearly unraveled. Different approaches for improving estimation accuracy are outlined. As an illustration, one of the solution strategies is implemented, which leads to accurate estimates of the breathing effort and relevant lung parameters.

In a society where digital data has become ubiquitous and has been projected to continue in this trajectory for the foreseeable future, machine learning has become a dependable tool to aid in analyzing these big datasets. However, where the data or machine learning algorithms are considered to be privacy-sensitive, one is then faced with the challenge of preserving the utility of machine learning in a privacy-preserving setting. In this paper, we focus on a use case where decentralized parties have privately owned machine learning algorithms, and would want to jointly generate a public model while not violating the privacy of their individual models, and data. We present ECONoMy: a privacy-preserving protocol that supports collaborative learning using an ensemble technique. Set in an honest-but-curious security model, ECONoMy is lightweight and provides efficiency and privacy in settings with large participant such as with IoT devices.