Skip to main content
Fig. 3 | BMC Medical Informatics and Decision Making

Fig. 3

From: Digital health system for personalised COPD long-term management

Fig. 3

An illustration of how the alerting thresholds were computed for the symptom diary score (a) and SpO2 data (b) from one patient. From the top anti-clockwise: (1) raw data points, (2) distribution of the data points, (3) probability density function (PDF), (4) cumulative density function (CDF). The threshold value is indicated by the small square in the last plot and it is equal to the value corresponding to the 95th percentile (5th in the case of SpO2). Gaussian kernels were used instead of rectangular ones in order to achieve a smooth PDF and CDF

Back to article page