Continence care has become a major care demand in nursing homes as over 50\% of their residents experience incontinence, i.e., the involuntary leakage of urine or stool [1]. In addition, it is time consuming and costly for the long-term care settings, and negatively impacts the residents' quality of life.
Nursing homes are recommended to implement continence assessments to improve continence care [2]. To this end, care personnel is requested to annotate continence care events on scoring sheets to define resident-specific continence profiles. Hence, another task for an already overburdened staff, resulting in forgotten entries and mistakes.
Many wearable (medical) devices found their way to the healthcare domain with an opportunity to support continence care management in nursing homes as well [3].
In this study, it was examined if DFree, a wearable device to predict toilet timing, could enrich the manual annotations from care personnel to draw up continence profiles and hence, support continence care management in nursing homes.

We evaluated the Toilet Timing Predicting Device DFree, a wearable from Triple W to detect the bladder size via non-invasive ultrasound technology. Seven nursing home residents over two nursing homes were recruited for participation to this exploratory field study. The sensor of DFree was placed at the abdomen of the participant in the area of the pubic bone. Data was collected for three to five days per participant, assisted by annotation of continence care events by care personnel through available nursing home technology. 
The Nurse Call System (NCS), Call Manager from Televic, is a communication and management application and was used by residents to trigger a call for assistance. If a resident called, care personnel could first select the reason of the call from a list and then an appropriate action to assist from a contextual list. A second technology, the Resident Care File (RCF), Geracc from Colilus, is a software application that stores information about the health and care plan of residents. For the study, additional tasks were implemented, ranging from toilet visits to the check up of incontinence material. At last, Texible Wisbi, a smart sensor mat  was placed between the mattress and sheet of the care bed for occupation detection.

During the field study, the researchers could monitor the DFree measurement through a web portal, facilitating interventions to recharge the device or reapply the sensor when applied incorrect or when the battery died. Based on this parallel monitoring, the participation from the last 2 residents was prematurely terminated as the observed measurement was incorrect. Triple W indicated that these poor measurements were most probably related to the presence of thick subcutaneous fat or large sagging skin, complicating an ultrasound measurement.
After the study, Triple W provided the researchers with the collected data for further analysis. Time series plots of all data obtained are visualised in Figure 1 for one resident, a 85 years old woman who suffered from functional incontinence (i.e., unable to get to a bathroom for one or more physical or mental reasons). Over the received raw bladder volume data of DFree (light purple area), a low-pass filter was applied for noise reduction (dark purple line). The occupation detector mat comprises information about the resting night times of the resident (brown). The NCS annotations (green) provide information about assistance on the toilet. At last, the annotations in the RCF give insight in whether a toilet visit was successful (blue) and the saturation level of the incontinence material (IM) in combination with a possible replacement (orange).

When looking at Figure 1, several patterns can be seen in the continence care of the resident. During the night, 'Occupied' is marked True and the resident is expected to be asleep, so little annotations in the NCS or RCF were expected. A morning always started with a successful toilet visit and the replacement of a highly saturated incontinence material, worn by the resident throughout the night. During the day, the resident was assisted for a toilet visit about three times. For some visits, she used the NCS to request care personnel to assist her from the toilet. For other visit, care personnel most likely entered the room before the resident meant to call them herself. When care personnel noticed the incontinence material was hardly saturated during a toilet visit, it was not yet replaced.
Figure 2 zooms in on the first part of Figure 1 to pair annotations with the bladder volume signal. Early in the second morning [A], one annotation was made in the RCF of a successful toilet visit in combination with the replacement of a highly saturated incontinence material. Over night, at least five significant drops in the bladder volume are depicted, which can explain the highly saturated incontinence material in the morning. One annotation was made via the NCS [B] to help the resident from the toilet after a successful toilet visit. This continence care event was also annotated in the RCF. As annotations in the RCF and NCS were made after providing care, it is expected that the voiding took place beforehand. Preceding the annotation, a decrease in bladder volume is noticeable. During the first day [C], three additional toilet visits were annotated in the RCF, which can all be associated with a decrease in bladder volume at the same period.

Patterns were found in the continence care of the resident, illustrating the added value of continence assessments and resident-specific continence profiles in nursing homes. Current practices, however, require manual annotation by already overburdened care personnel. In this study, we aimed to investigate whether DFree, a toilet timing predicting wearable, could enrich these annotations and support continence care management. 
DFree measured seven participants' voiding times and provided a bladder volume signal. This signal was additionally filtered to reduce noise. Several annotations in the NCS and RCF could be linked with a decrease in the signal, representing a voiding time. Therefore, we will continue our research with an in-depth analysis of the data to scientifically evaluate the reliability of the device.

Offermans, M. P., Du Moulin, M. F., Hamers, J. P., Dassen, T., & Halfens, R. J. (2009). Prevalence of urinary incontinence and associated risk factors in nursing home residents: a systematic review. Neurourology and Urodynamics: Official Journal of the International Continence Society, 28(4), 288-294.
Wagg, A., Kung Chen, L., Johnson II, T., Kirschner-Hermanns, R., Kuchel, G., Markland, A., ... & Wyman, J. (2017). Incontinence in frail older persons. Incontinence: 6th International Consultation on Incontinence, Tokyo, September 2016, 1309-1442.
Acampora, G., Cook, D. J., Rashidi, P., & Vasilakos, A. V. (2013). A survey on ambient intelligence in healthcare. Proceedings of the IEEE, 101(12), 2470-2494.