EP2194515A1

EP2194515A1 - Subject monitoring

Info

Publication number: EP2194515A1
Application number: EP08170465A
Authority: EP
Inventors: designation of the inventor has not yet been filed The
Original assignee: Philips Intellectual Property and Standards GmbH; Koninklijke Philips Electronics NV
Current assignee: Philips Intellectual Property and Standards GmbH; Koninklijke Philips NV
Priority date: 2008-12-02
Filing date: 2008-12-02
Publication date: 2010-06-09

Abstract

A system for monitoring one or more subjects in respective units such as beds comprises a first device, such as a display device arranged to present a plurality of categories, an input device arranged to receive an input for a subject, the input selecting a category for the subject, one or more sensors arranged to monitor the or each subject, for which a category has been selected, a processing device arranged to receive an output from the or each sensor, the output indicating movement of a subject with respect to their respective bed, and to execute an algorithm using the selected category and the received output, and an output device arranged to generate an output signal such as an alarm signal, according to the executed algorithm.

Description

TECHNICAL FIELD OF THE INVENTION

This invention relates to a method of, and system for, monitoring one or more subjects in respective units such as beds. In one embodiment, the invention provides a notification scheme for bed departures of agile and fall prone residents, for example, for use in nursing homes.

BACKGROUND OF THE INVENTION

Generally, carers in nursing homes want to be informed when the elderly in their care leave their bed at night. Depending on the nature of the frailty of the elderly subjects, the nurses want to be informed sooner or later about the bed departure. For instance, the carers would like to be informed even before the bed departure of a fragile elderly in order to avoid falls. On the other hand, it is sufficient to notify the carer if an agile but elderly subject has not returned to bed after a couple of minutes, since this could indicate that the subject has left their room and is roaming in the nursing home. No notification and hence no additional workload for the nurses is necessary if the agile elderly subject has returned to bed within some minutes, though.
Current algorithms for presence detection are complicated and depend on several parameters. In order to achieve a desired alarming scheme for the individual subjects that are being monitored, for example, in elderly homes, these parameters have to be tuned. Such a tuning is too complex to be carried out by the nurses, though. In large nursing homes, the carers will receive numerous bed departure notifications during a single night. Given the high workload of the nurses they will not be able to react to every notification immediately.

SUMMARY OF THE INVENTION

It is therefore an object of the invention to improve upon the known art.
According to a first aspect of the present invention, there is provided a method of monitoring one or more subjects in respective units comprising presenting a plurality of categories, at a first device, receiving an input for a subject, the input selecting a category for the subject, at an input device, monitoring the or each subject, for which a category has been selected, with a respective sensor, receiving an output from the or each sensor, the output indicating movement of a subject with respect to their respective unit, at a processing device, executing an algorithm using the selected category and the received output, at the processing device, and generating an output signal, according to the executed algorithm, at an output device.
According to a second aspect of the present invention, there is provided a system for monitoring one or more subjects in respective units comprising a first device arranged to present a plurality of categories, an input device arranged to receive an input for a subject, the input selecting a category for the subject, one or more sensors arranged to monitor the or each subject, for which a category has been selected, a processing device arranged to receive an output from the or each sensor, the output indicating movement of a subject with respect to their respective unit, and to execute an algorithm using the selected category and the received output, and an output device arranged to generate an output signal, according to the executed algorithm.
Owing to the invention, it is possible to provide a simple subject monitoring system that does not place a complex workload on staff such as nurses and carers, who generally do not have the technical skills to work current systems satisfactorily. With the present solution the nurses only have to assign the elderly to one of the risk groups whereupon the parameters are set to suitable values automatically. There is provided a method in which the notification via an output (alarm) signal can be organized hierarchically such that the nurses get a clear overview of which resident needs their assistance most urgently.
As mentioned above, carers in nursing homes are generally interested in getting informed when the elderly in their care leave their bed at night. For some elderly (for example fall prone subjects) these notifications should be sent as early as possible while for others a notification is only necessary if they have not returned to bed after several minutes. In order to obtain the desired notification scheme for the individual types of elderly numerous parameters have to be tuned in the algorithm used to detect the presence in bed. In current systems, such a tuning is too demanding for the carers, though. With the solution delivered by this invention the carers only have to assign the elderly to risk groups while the suitable parameter settings are selected automatically. Since caretakers may receive numerous bed departure notifications in large nursing homes there is also presented a scheme with which these messages can be organized hierarchically.
In this invention there is provided a method with which the desired notification scheme for the bed departure of an elderly can be activated by assigning the subject to a risk group. The parameters of the presence detection algorithm are then set automatically to predefined values. The ordering scheme provides that the numerous notifications to the carers are organized hierarchically by the risk groups the monitored people belong to and/or chronologically. The messages disappear once the monitored persons have returned to bed.
Preferably, the execution of the algorithm in respect of the received output detects a predefined event. The predefined event comprises, for example whether a subject has sat up in bed, whether a subject has left their bed or whether a subject has returned to their bed. The output of a sensor that is monitoring a subject has to be processed to determine a meaningful fact (within a certain margin of error) about the subject. The output can be converted into events, such as a subject leaving their bed, and this provides a simple and efficient method of then determining whether to generate an output signal such as an alarm call, based upon the risk group to which the subject has been assigned.
Advantageously, the method further comprises accessing a predetermined time period for the selected category, for use by the algorithm, wherein the output signal is generated when, following detection of a predefined event, the predetermined time period has expired. This step develops the event idea further, and means for example, that a subject in a first group of subjects (very infirm, for example) will have an alarm generated very soon after they leave their bed, or indeed before they have even left their bed, whereas a subject in a second group of subjects (physically fit, but suffering from dementia, for example) will have an alarm generated perhaps two minutes after they leave their bed. This ensures that the parameters can be expressed in simple terms of time periods, and can be easily adjusted by a technician if needed.

BRIEF DESCRIPTION OF THE DRAWINGS

Embodiments of the present invention will now be described, by way of example only, with reference to the accompanying drawings, in which:-

Fig. 1 is a schematic diagram of a system for monitoring subjects in their beds,
Fig. 2 is a diagram showing force levels detected by a sensor of the system of Fig. 1, also showing autocorrelation and variance of the force levels,
Fig. 3 is a schematic diagram of a display device of the system of Fig. 1,
Fig. 4 is a schematic diagram of a second embodiment of a monitoring station for use in the system of Fig. 1,
Fig. 5 is a flowchart of a method of monitoring subjects, and
Fig. 6 is a table of parameters.

DETAILED DESCRIPTION OF THE EMBODIMENTS

Shown schematically (not to scale) in Fig. 1 is a system for monitoring one or more subjects in units 10. A unit 10 can be anything that a person can rest on (preferably lying on with their entire body, and preferably not touching the ground with their feet), for example beds, couches, certain forms of seats to rest on etc. In Fig. 1, the units 10 are shown as beds 10. Only four beds 10 are shown, but the system can be extended to cover a much larger number of beds 10, as desired. The beds 10 could be located in a nursing home, where elderly subjects are cared for, who are physically and/or mentally infirm. In general, in such establishments, a number of carers/nurses will look after the subjects, where the number of carers is significantly less than the number of subjects. This is particularly true at night, where it is assumed that a lower demand will be placed on the nurses during this period.
In order to assist the monitoring of the subjects by the carers, it is desirable to have an automatic system for generating an alarm in relation to the movements of the subjects, particularly at night. This automatic monitoring is provided by the system shown in Fig. 1. Each bed 10 is provided with a respective sensor 12. This sensor 12 can be a pressure sensor 12, for example, which provides a reading in relation to the occupancy and position of the subject who uses the respective bed 10. It had been found that the best position for the sensor 12 is at a placement that would approximate to the chest position of the subject in the bed 10. An alternative embodiment for the sensor 12 used in the bed departure detection can be a strain gauge. Adhesive can be used to glue the strain gauge onto a slat of a slatted bed frame which is approximately located under the thorax of the subject. When the subject is in bed the slat is bent, i.e. becomes longer and this is a physical change that the strain gauge actually detects.
Each sensor 12 is connected via a network 14, which could be wired or wireless in a conventional fashion, to a monitoring station 16. In Fig. 1, this monitoring station is shown as a conventional computer 16, but a dedicated unit 16 could also be provided to accomplish the same task. The monitoring station 16 comprises a first device, being a display device 18, a processing device 20 and an input device 22, being (in this embodiment) a computer keyboard. An alternative input device 22 could be a touchscreen, discussed below with reference to Fig. 3. Information from the sensors 12 is passed directly to the monitoring station 16 in real-time and processed there by the processing device 20. According to the processing performed by the monitoring station 16, then an alarm may be sounded to alert a carer that potentially a subject needs assistance. In the embodiment of Fig. 1, the processing is carried out remotely at the monitoring station 16, but all of the processing could be carried out locally at the bed 10, with individual processing stations at each bed 10.
Fig. 2 shows at plots a) and d) the typical force pattern recorded with the sensor 12 when a person leaves their bed 10. Plot b) shows the autocorrelation coefficient of the signal shown in a) and d). Plot c) shows the variance of the signal shown in a) and d). Plots a) and d) show the typical signal recorded by a strain gauge sensor 12 when a person is leaving their bed 10. The parts of the signal indicated by the arrows 24 show the periods of time during which the nurses should help the elderly subject. In plot a) this is shown for a fall prone resident while in plot b) the alerting scheme for an agile person is depicted.
In plot b) the autocorrelation coefficient of the signal is plotted. For this purpose a moving window of length L is used and in each window the autocorrelation coefficient (delay one data point) of the signal was computed. In plot c) the variance of the signal, also estimated in moving windows of size L is shown.
In the beginning of the signal shown in Fig. 2 at the plots a) and d) the monitored person is lying in their bed 10 such that force detected by the respective sensor 12 is comparatively high. At point 1) the person sits up in bed and stays on the bed edge until point 3). During this period the amplitude of the signal is already low, however, there are still some smaller fluctuations visible which appear as the person is still in contact with the mattress. These fluctuations lead to a considerable autocorrelation coefficient, as shown in plot b), which becomes very small, however, once the person eventually leaves the bed at point 3).
The period of time between point 1) and 3) can be used by the carers to reach the room of a fall prone elderly and assist before they have left their bed 10. Hence, a notification is sent to the nurse once the amplitude of the signal has fallen under the threshold a_out ^fragile for more than τ_out ^fragile seconds. Note, that the alarm is not sent immediately (τ_out ^fragile =0), in order to avoid false alarms which could occur, for example, when a subject is turning around in their bed 10. A false alarm could still happen, though, if the elderly subject lies down again and does not leave the bed.
So far the monitoring of the subject has been described using a pressure sensor 12, but it should be mentioned that in other embodiments different strategies can be used to detect the event of the subject getting out of bed. In particular, vital body signs (such as heart rate, heart rate variability or breathing rate) can be used to increase the period in which the bed departure event can be predicted.
In order to avoid false alarms, the nurses are notified at point 3) if an agile wanderer is monitored. At this point in time, the person has eventually left the bed such that the carers should be notified. Algorithmically, this means that the notification is sent if both the signal amplitude stays under a_out ^fragile and if the autocorrelation coefficient is smaller than xcor_out for more than τ_out ^wander seconds. Since in the case of agile persons no false alarms are more important than a very early notification of the nurses, a_out ^agile < a_out ^fragile and τ_out ^wander > τ_out ^fragile are used.
For agile people who usually do not wander at night no departure notification needs to be sent if they leave their bed for a couple of minutes only (for example, if they just want to go to the toilet). If they stay out of bed for longer periods of time then carers will still appreciate a notification as it could mean that the subject is in need of help. Seen from the algorithmic perspective this means that the same criteria are used as for wanderers, however, a longer time threshold τ_out is used.
Similar considerations hold for the cancellation of the bed departure notification. For a fall prone person the alarm should only be cancelled if the person has lain down calmly since a high activity in bed often precedes a further bed departure. Accordingly, the notification is cancelled only if the signal amplitude is larger than the threshold a_in for more than τ_in seconds and if the variance of the signal is smaller than var_in. This is the case at point 2) in Fig. 2. For an agile person, it is enough to know that they have lain down in bed, which is when the signal amplitude is larger than a_in for more than τ_in seconds, to cancel the bed departure notification, as shown at point 4 in Fig. 2.
In order to characterise the subj ects in their care, the nurses are presented with a plurality of categories, at the first device, being the display device 18, and the nurse will enter a user input for a subj ect, the user input selecting a category for the subj ect, at the user input device 22. This categorisation of the subjects can be done at any time, and the category assigned to a subject can also be changed at any time, according to any perceived change in the condition of a subject. The categories are designed to be based upon easily recognisable physical traits, so that a relatively low technically skilled member of staff can make the necessary selection.
The first device 18, as discussed above, can provide information to the nurses on the displays of pagers and phones. It is also possible that audible-only signals could be provided to the nurses. In this case, a loudspeaker will present the nurse with a set of categories, and a microphone with speech recognition software can be used by the nurse to input the selected category.
Fig. 3 illustrates a screenshot of a graphical user interface to the bed notification screen. The display device 18 is showing a window 26 which allows a user such as a nurse or carer to make a selection regarding the group to which the specific subject belongs. Three categories 28 are shown, being fall prone, agile and agile + dementia. These categories 28 are shown to illustrate the concept of the invention, more or fewer categories may be used, depending upon the specific monitoring situation. The subject 30 is shown with their name and bed number and a photograph, in order to assist the nurse in the subject identification and selection. In this embodiment, the nurse simply has to click onto a category 28 in order to select a category 28 for the subject 30, at the user input device 22. The input for selecting the category of the resident could be done automatically, for example using the output of another patient classification system (Patient Database of Hospitals or healthcare systems) as an input to the system.
In the above paragraph, three possible risk groups are mentioned (fall prone, agile, agile + dementia). The risk that somebody is prone to fall can be assessed by means of a predefined checklist. The result of such an assessment is a score which reflects the risk of falling. In the user interface there is provided a field into which the score of the fall risk assessment can be entered. Based on that score, the parameters of the algorithm are adjusted automatically. In this way the selection of category is carried out indirectly based on answers to a checklist.
Once the category 28 has been selected for one or more subjects 30, then the system of Fig. 1 is monitoring each subject 30, for which a category 28 has been selected, with a respective sensor 12. As detailed above, there is received an output from each sensor 12, the output indicating movement of a subject 30 with respect to their respective bed 10, at the processing device 20. The processing device 20 executes an algorithm using the selected category 28 and the received output, and generates an output signal such as an alarm signal, according to the executed algorithm, at an output device. This alarm signal can differ depending upon the nature of the movement detection and the category concerned. For example a fall prone subject leaving their bed 10 can be seen to be a more urgent situation than an agile + dementia subject who has not yet returned to their bed.
Fig. 4 shows an example of alternative embodiment of the monitoring station 16. In this case the station 16 is a purpose built unit that connects to the network 14. The unit includes a touchscreen 32, which functions both as the display device and the input device for the presenting and input of information to and from the nurses and carers. The monitoring station 16 is also equipped with dedicated output devices 34, being a warning light 34a and a loudspeaker 34b. The output devices 34 together will provide an output signal, if the processing function of the station 16 indicates that a situation has arisen in relation to a specific subject. It is also possible to use paging systems which are connected to the alarming device 34, to notify nurses about the existence of an alarm signal. Further, it is possible to send the alerts to a cordless device (for example DECT telephones) that nurses commonly use.
As can be seen in Fig. 4, when an alarm is generated, then the touchscreen 32 will display details of the specific subject 30 for whom the alarm has been generated, along with their name and bed location, as well as the category to which that subject 30 has earlier been assigned. The carers can see immediately the group to which the subject 30 belongs, and can tailor their action accordingly. Any changes to the current status of the subject 30 can also be reflected by the touchscreen 32. For example, if the subject 30 returns to their bed 10, which results in the alarm for that subject 30 being cancelled, then this can be displayed to the carers, via the touchscreen 32.
The approach described above of monitoring the subjects 30 is summarised in the flowchart shown in Fig. 5. The appropriate values for the mentioned parameters are shown in Table 1, shown in Fig. 6. Table 1 shows preferable parameter sets for the individual notification schemes. Fig. 5 shows a sample flowchart of the preferred algorithm. In the first column, the parameters are set and the samples are read in. The second column is related to the bed departure notifications while the third column handles the cancellation of them. As detailed above, it is difficult for a nurse to tune the above mentioned parameters in order to achieve the desired notification scheme. For this reason the carers just have to assign the elderly subjects to one of the above mentioned categories (for example, fragile, wanderer, agile), whereupon the appropriate parameters are set automatically.
The execution of the algorithm in respect of the received output detects a predefined event. The predefined event can comprise whether a subject has sat up in bed, whether a subject has left their bed or whether a subject has returned to their bed. The algorithm is operated to access a predetermined time period for the selected category, present in the table of Fig. 6, wherein the alarm signal is generated when, following detection of a predefined event, the predetermined time period has expired.
The sample algorithm, shown in Fig. 5, will notify the carers of situations when they should attend to a subject. In larger nursing homes, where several people are monitored simultaneously, the carers will receive numerous notifications. It is advantageous to list the notifications generated by the algorithm in the following order. First the notifications of fall prone people are shown in chronological order as they need the assistance of the nurses most urgently. Then the notifications originating from the wanderers appear chronologically. Finally, the notifications of agile residents are shown in chronological order. The messages disappear from the list once the respective elderly subject has returned to their bed.

Claims

A method of monitoring one or more subjects (30) in respective units (10) comprising:
- presenting a plurality of categories (28), at a first device (18; 32),

- receiving an input for a subject (30), the input selecting a category (28) for the subject (30), at an input device (22; 32),

- monitoring the or each subject (30), for which a category (28) has been selected, with a respective sensor (12),

- receiving an output from the or each sensor (12), the output indicating movement of a subject (30) with respect to their respective unit (10), at a processing device (20),

- executing an algorithm using the selected category (28) and the received output, at the processing device (20), and

- generating an output signal, according to the executed algorithm, at an output device (34).
A method according to claim 1, wherein the execution of the algorithm in respect of the received output detects a predefined event.
A method according to claim 2, wherein unit (10) comprises a bed (10), and the predefined event comprises whether a subject (30) has sat up in bed (10), whether a subject (30) has left their bed (10) or whether a subject (30) has returned to their bed (10).
A method according to claim 2 or 3, and further comprising accessing a predetermined time period for the selected category (28), for use by the algorithm, wherein the output signal is generated when, following detection of a predefined event, the predetermined time period has expired.
A method according to any preceding claim, and further comprising setting parameters for the algorithm, according to the selected category (28).
A system for monitoring one or more subjects (30) in respective units (10) comprising:
- a first device (18; 32) arranged to present a plurality of categories (28), an input device (22; 32) arranged to receive an input for a subject (30), the input selecting a category (28) for the subject (30),

- one or more sensors (12) arranged to monitor the or each subject (30), for which a category (28) has been selected,

- a processing device (20) arranged to receive an output from the or each sensor (12), the output indicating movement of a subject (30) with respect to their respective unit (10), and to execute an algorithm using the selected category (28) and the received output, and

- an output device (34) arranged to generate an output signal, according to the executed algorithm.
A system according to claim 6, wherein the processing device (20) is arranged, when executing the algorithm in respect of the received output, to detect a predefined event.
A system according to claim 7, wherein unit (10) comprises a bed (10), and the predefined event comprises whether a subject (30) has sat up in bed (10), whether a subject (30) has left their bed (10) or whether a subject (30) has returned to their bed (10).
A system according to claim 7 or 8, wherein the processing device (20) is further arranged to access a predetermined time period for the selected category (28), for use by the algorithm, and wherein the output device (34) is arranged to generate the output signal when, following detection of a predefined event, the predetermined time period has expired.
A system according to any one of claims 6 to 9, wherein the processing device (20) is further arranged to set parameters for the algorithm, according to the selected category (28).