Fig 1.
An unsupervised framework for early UTI prediction (left) a visualisation of six-hour Sensor Firing Pattern (SFP) data matrix (right).
Table 1.
Description of TIHM daily health score, adapted from NEWS2 [55].
Table 2.
Performance of supervised and unsupervised UTI detection models.
Fig 2.
Cluster categorisation of P1 SFPs (top); visualisation of six-hour SFP data matrix belonging to HSFP category (middle) and visualisation of six-hour SFP data matrix belonging to RSFP category (below).
Table 3.
Confusion matrix (left) and numerical evaluation (right) of automated sleep analysis (ASA) algorithm vs. self-reports (SR) for 28 participants.
Fig 3.
Demonstration of aggregated data collected from an individual’s home for two test days, a normal day (top) versus an abnormal one (middle) and their corresponding night-time sleep pattern (below).
The data is aggregated in one hour interval and normalised to ensure that the activity level of each sensor is ranged between 0 to 1.