- Ponencia en CongresoLPV control of glucose for diabetes type I(2010)"This paper considers the problem of automatically controlling the glucose level in a Diabetes type I patient. Three issues have been considered: model uncertainty, timevarying/nonlinear phenomena and controller implementation. To that end, the dynamical model of the insulin/glucose relation is framed as a Linear Parameter Varying system and a controller is designed based on it. In addition, this framework allows not only a better performance than other classical methods, but also provides stability and performance guarantees. Design computations are based on convex Linear Matrix Inequality (LMI) optimization. Implementation is based on a low order controller whose dynamics adapts according to the glucose levels measured in real-time."
- Ponencia en CongresoControl-oriented linear parameter-varying model for glucose control in type 1 diabetes(2016)"The contribution of this paper is a controller design oriented model of insulin-glucose dynamics in Type 1 Diabetes Mellitus (T1DM). The novelty of the proposed model is to more effectively include the time-varying nature, and also the inter-patient variability, associated with the glucose control problem. Importantly, this is achieved in a manner that straightforwardly facilitates well-known and standard controller synthesis procedures. In that way, an average Linear Parameter-Varying (LPV) model that captures the dynamics from the insulin delivery input to the subcutaneous-glucose concentration output is constructed based on the Universities of Virginia (UVA)/Padova metabolic simulator. In addition, a system-oriented reinterpretation of the classical ad-hoc 1800 rule is applied to adapt the model’s gain."
- Ponencia en CongresoTwo-dimensional posture evaluation in Parkinson’s disease: effect of loads on the spinal angle during gait(2016)"Parkinson’s Disease patients present diminished coordination caused by neural degeneration. This leads to large motor difficulties during gait such as balance loss and pronounced forward inclination of the upper body. This work assessed the spinal sagittal plane angle alterations in two groups: six parkinsonian patients and six control healthy subjects. This parameter was analyzed during gait under three conditions: without external loads and with external loads applied either on the chest or on the lower back area. Results were statistically compared by means of t-test of paired samples in both groups. For patients, a significant effect was found when loads were applied on the chest. On the other hand, healthy subjects showed no significant differences in either case."
- Ponencia en CongresoAutomatic detection of reverse‑triggering related asynchronies during mechanical ventilation in ARDS patients using flow and pressure signals(2019)"Asynchrony due to reverse-triggering (RT) may appear in ARDS patients. The objective of this study is to validate an algo-rithm developed to detect these alterations in patient–ventilator interaction. We developed an algorithm that uses flow and airway pressure signals to classify breaths as normal, RT with or without breath stacking (BS) and patient initiated double-triggering (DT). The diagnostic performance of the algorithm was validated using two datasets of breaths, that are classified as stated above. The first dataset classification was based on visual inspection of esophageal pressure (Pes) signal from 699 breaths recorded from 11 ARDS patients. The other classification was obtained by vote of a group of 7 experts (2 physicians and 5 respiratory therapists, who were trained in ICU), who evaluated 1881 breaths gathered from recordings from 99 sub-jects. Experts used airway pressure and flow signals for breaths classification. The RT with or without BS represented 19% and 37% of breaths in Pes dataset while their frequency in the expert’s dataset were 3% and 12%, respectively. The DT was very infrequent in both datasets. Algorithm classification accuracy was 0.92 (95% CI 0.89–0.94, P < 0.001) and 0.96 (95% CI 0.95–0.97, P < 0.001), in comparison with Pes and experts’ opinion. Kappa statistics were 0.86 and 0.84, respectively. The algorithm precision, sensitivity and specificity for individual asynchronies were excellent. The algorithm yields an excellent accuracy for detecting clinically relevant asynchronies related to RT."
- Ponencia en CongresoEpilepsy seizure onset detection applying 1-NN classifier based on statistical parameters(2018)"Epilepsy is a disease caused by an excessive discharge of a group of neurons in the cerebral cortex. Extracting this information using EEG signals is an ongoing challenge in biomedical signal processing. In this paper, a new method is proposed for onset seizure detection in epileptic EEG signals based on parameters from the t-location-scale distribution coupled with the variance and the Pearson correlation coefficient. The 1-nearest neighbor classifier achieved a 91% sensitivity (True positive rate) and 95% specificity (True Negative Rate) with a delay of 4.5 seconds (on average) in the 45 signals analyzed, which suggests that the proposed methodology is potentially useful for seizure onset detection in epileptic EEG signals."