A novel intelligent portable home oxygen therapy unit for patients with chronic respiratory failure

Portable oxygen concentrators (POCs) allow patients to carry out normal life outside the home. In existing POCs, patients must regulate, manually and continuously, the O2 flow, in accordance with the intensity of the physical activity (PA). This manual operation requires patient’s attention and training, and is often subject to errors that can induce hypoxia or hyperoxia. This study aimed at developing and evaluating a novel intelligent portable oxygen concentrator (iPOC), that automatically identifies the intensity of patient’s PA level and, autonomously, adjusts the O2 flow to the level prescribed by the clinician. An external portable electronic system was designed and integrated into two commercially available POCs. The system comprised two units. First, a sensor module attached to the patient that classified the PA every 3 seconds into three categories: sedentary (sitting, standing still, lying down), light (walking) and moderate (going up or down stairs or ramps). Second, a receiver unit, located next to the POC, that automatically adjusted the O2 flow according to the PA detected in real-time. No user intervention was necessary. The algorithms were designed with data from 8 subjects. These subjects walked a predefined circuit for 20 minutes. The circuit included walking- flat areas, ramps and staircases. Artificial intelligence techniques were used for the automatic classification of PA. The achieved accuracy, defined as the proportion of 3-second periods correctly identified among the total number of periods of the experiment was 93%. Sensitivity and specificity values were 92% and 94% respectively. The proposed solution can support the optimization of ambulatory oxygen therapy.