WO2017207020A1

WO2017207020A1 - Compliance metric for the usage of hygiene equipment

Info

Publication number: WO2017207020A1
Application number: PCT/EP2016/062156
Authority: WO
Inventors: Håkan Lindström
Original assignee: Sca Hygiene Products Ab
Priority date: 2016-05-30
Filing date: 2016-05-30
Publication date: 2017-12-07

Abstract

Estimating a compliance metric indicating the usage of hygiene equipment by one or more operators, using a first receiving section configured to receive usage data from an equipment sensor arrangement, said usage data indicating a usage of said hygiene equipment; a second receiving section configured to receive first opportunity data from a first opportunity sensor arrangement, said first opportunity data indicating a first set of opportunities to use said hygiene equipment, and configured to receive second opportunity data from a second opportunity sensor arrangement, said second opportunity data indicating a second set of opportunities to use said hygiene equipment, wherein the second opportunity sensor arrangement is reduced relative to said first opportunity sensor arrangement; and a calculation section configured to estimate, based on said usage data and said first opportunity data, a function for estimating said compliance metric from said usage data and said second opportunity data.

Description

COMPLIANCE METRIC FOR THE USAGE OF HYGIENE EQUIPMENT

Technical Field

The present disclosure generally relates to estimating a compliance metric in the context of hygiene equipment, such as soap, disinfectant, and/or towel dispensers, and the like. More particularly, the present disclosure relates to ways of determining a value of compliance metric estimate that indicates the actual usage of hygiene equipment by corresponding sensors.

Background

Hygiene equipment is commonplace today in many facilities, such as hospitals, medical service centers, intensive care units, day clinics, private practices, lavatories, rest rooms, hotels, restaurants, cafes, food service places, schools, kindergartens, manufacturing sites, administration and office buildings, and, in general, places and facilities that are accessible to the public or to a considerable number of individuals. The mentioned hygiene equipment thereby includes various types of individual devices and installations such as soap dispensers, dispensers for disinfectant solutions, gels or substances, towel dispensers, glove dispensers, tissue dispensers, hand dryers, sinks, radiation assisted disinfectant points, ultraviolet (UV) light, and the like.

Although such hygiene equipment is commonplace today in many places, the use thereof by the individuals visiting these places or working in these places is still oftentimes not satisfactory. For example, hospitals, and, in general, medical service centers often suffer from hygiene deficiencies, which, in turn, may lead to the spread of infections and related diseases. In particular, such insufficient hygiene amongst medical care personnel coming into close contact with patients and bodily fluids can lead to the spread of infectious diseases amongst the personnel and other patients. It is also known that infections by highly resistant bacteria pose a severe problem in such places, especially in hospitals. In general, so-called Healthcare Associated Infections (HAI) are a real and tangible global problem in today's healthcare. HAI can be found to be currently the primary cause of death for 140.000 patients/year, affecting millions more and costs society in the range of billions of EUR/year.

At the same time, however, it is known that hygiene, and, in particular, hand hygiene, is an important factor as far as the spread of infectious diseases are concerned. Specifically, medical care personnel should make proper use of hand hygiene as often as possible so that the spread of bacteria and other disease causing substances is minimized. The actual usage of such hygiene equipment, however, may depend on - amongst others - the management of the facility, accessibility and usability of the equipment, culture, the cooperation and will exercised by the individuals working in these places or visiting such places, training of individuals, time pressure, and possibly also other factors. In other words, an important factor remains the fact that individuals may not make use of installed and provided hygiene equipment although they are supposed to. Furthermore, it is generally accepted that an increased use of hygiene equipment can substantially contribute in reducing the spread of bacteria and the like, which, in turn, can drastically reduce the appearance of related infections and diseases. As a consequence, one may have considerable interest in a so- called compliance that in some way or another compares the actual use of hygiene equipment to some sort of target usage. For example, a corresponding relatively low compliance metric may indicate that the actual use of hygiene equipment is not satisfactory, whilst relatively high compliance metric may indicate that the actual use of hygiene equipment corresponds, within a given threshold, to some target usage, and, consequently, may be regarded as being satisfactory. Such a compliance metric may provide many advantages, since it gives a concise picture to operators of the corresponding facility so that they may initiate measures for increasing the actual use of hygiene equipment.

Therefore, there are already ways of estimating such a compliance metric in the arts, wherein the conventional approaches usually rely on measuring and/or observe "manually" by a human observer so-called opportunities and comparing these obtained opportunities to a measured/detected/observed actual use of the hygiene equipment. In other words, the opportunities indicate any event when hygiene equipment should or could have been used. By then comparing this "should/could"-value to an actual usage value a compliance metric can be calculated, as e.g. a percentage value or a ratio. In general, the opportunities can be well defined figures, since they may be associated to specific rules and/or recommendations. For example, the World Health Organization (WHO) has defined the so-called "Five Moments Of Hand Hygiene" (cf. www.who.int/psc/tools/ Five_moments/en/ ) including as explicit definitions for opportunities: 1. Before patient contact; 2. Before aseptic task; 3. After body fluid exposure risk; 4. After patient contact; and 5. After contact with patient surroundings. Moreover, measurements on a corresponding hand hygiene compliance is becoming a regulatory requirement for the healthcare sector and may serve as an important quality improvement tool. In this context it should be noted that, whilst it is commonplace to implement sensor arrangements in hygiene equipment for measuring the actual usage, it may be more difficult to implement sensor arrangements for measuring, detecting, and/or sensing opportunities. The prior arts generally attempt to improve the sensor arrangement used for sensing/measuring the opportunities for, in turn, improving the accuracy of the obtained compliance metric. However, the optimization of such sensor arrangement for sensing/measuring opportunities in the context of hygiene equipment usage remains complex and difficult, since it may require additional and possibly expensive hardware, equipment, or resources. Moreover, surveillance equipment may also interfere with privacy regulations when corresponding opportunity sensor equipment comprises cameras, position tracking systems, or related image and data-processing systems. Thus, even though systems for hand hygiene compliance monitoring may exist, no global standard has been established. The global golden standard for hand hygiene compliance measurement is still to employ a manual observer that observes the hand hygiene practices for a certain time period and calculates the hand hygiene compliance rate based on this observational period.

There is therefore a need for an improved system and method of estimating a compliance metric that improves accuracy of the measured compliance, whilst being as little as possible dependent on complex, expensive, and/or undesired equipment or otherwise labor-intense surveillance and measurement techniques .

Summary

The mentioned problems and drawbacks are addressed by the subject matter of the independent claims. Further preferred embodiments are defined in the dependent claims. According to one aspect of the present invention there is provided a system for estimating a compliance metric indicating the usage of hygiene equipment by one or more operators, the system comprising a first receiving section configured to receive usage data from an equipment sensor arrangement, said usage data indicating a usage of said hygiene equipment; a second receiving section configured to receive first opportunity data from a first opportunity sensor arrangement, said first opportunity data indicating a first set of opportunities to use said hygiene equipment, and configured to receive second opportunity data from a second opportunity sensor arrangement, said second opportunity data indicating a second set of opportunities to use said hygiene equipment, wherein the second opportunity sensor arrangement is reduced relative to said first opportunity sensor arrangement; and a calculation section configured to determine, based on said usage data and said first opportunity data, a function for estimating said compliance metric from said usage data and said second opportunity data.

According to another aspect of the present invention there is provided a method for estimating a compliance metric indicating the usage of hygiene equipment by one or more operators, the method comprising: (a) receiving usage data from an equipment sensor arrangement, said usage data indicating a usage of said hygiene equipment; (b) receiving first opportunity data from a first opportunity sensor arrangement, said first opportunity data indicating a first set of opportunities to use said hygiene equipment; (c) determining, based on said usage data received in the first step and said first opportunity data, a function for estimating said compliance metric from usage data and second opportunity data; (d) receiving usage data from the equipment sensor arrangement; (e) receiving said second opportunity data from a second opportunity sensor arrangement, said second opportunity data indicating a second set of opportunities to use said hygiene equipment, wherein the second opportunity sensor arrangement is reduced relative to said first opportunity sensor arrangement; and (f) estimating said compliance metric from usage data received in the second step and said second opportunity data using said function.

Brief description of the drawings

Embodiments of the present invention, which are presented for better understanding the inventive concepts but which are not to be seen as limiting the invention, will now be described with reference to the figures in which:

Figure 1 shows a schematic view of a deployment of a system for estimating a compliance metric according to an embodiment of the present invention;

Figure 2 shows a schematic view of a deployment of a system for estimating a compliance metric according to another embodiment of the present invention;

Figures 3A and 3B show schematic views in conjunction with a neural network being used for determining a function according to a further embodiment of the present invention;

Figure 4A shows a flowchart of an exemplary way to determine parameters of a function according to an embodiment of the present invention; Figure 4B shows a flowchart of an exemplary way to employ the learned function according to an embodiment of the present invention;

Figure 4C shows a flowchart of a general method embodiment of the present invention; and

Figure 5 shows a schematic view of a general entity embodiment of the present invention .

Detailed Description

Figure 1 shows a schematic view of a deployment of a system for estimating a compliance metric according to an embodiment of the present invention. The system is generally deployed for estimating a compliance metric that indicates the usage of hygiene equipment 10, in the form of, for example, a number of individual pieces of hygiene equipment such as the shown soap or disinfectant dispensers 10-1, 10-2, and 10-N. Generally, the hygiene equipment 10 can comprise any one of a soap dispenser, a dispenser for disinfectant solutions, gels or substances, a towel dispenser, a glove dispenser, a tissue dispenser, a hand dryer, a sink, a tap, and a radiation assisted disinfectant point, a UV disinfecting apparatus and the like. Such equipment is generally deployed in a facility being any of a hospital, a medical service center, an intensive care unit, a day clinic, a private practice, a lavatory, a rest room, restaurants, cafes, food service places, schools, kindergartens, a hotel, a manufacturing site, an administration or office building, a shopping center, and, in general, any places and facility that is accessible to the public or to a considerable number of individuals . The system 30 comprises a first receiving section 31 that is configured to receive usage data from the hygiene equipment 10 by an equipment sensor arrangement 100. Said equipment sensor arrangement 100 is a collection of all available sensors that are able to generate and forward individual usage signals u-1, -2 , ... that indicate an actual use of some or a specific piece of hygiene equipment. For example, a soap dispenser 10-1 may be provided with a sensor 100-1 that is configured to generate a usage signal u-1 whenever an operator actually uses the piece of hygiene equipment and ejects an amount of soap. In this way, the first receiving section 31 receives the usage data in the form of individual signals u-1, u-2,... and may thus store the general usage data U as a whole as data 310 in a database 34.

The signals u-1, u-2,... are typically signaling "now it happened" (e.g. by carrying a Boolean value "TRUE" or by simply carrying data such as a dispenser ID) . In a way, the mere fact that a signal is received may indicate to the system 30 that a usage event happened. However, the signal may also include more information, including information on when the piece of hygiene equipment was used (e.g. timestamp) , where it was used (e.g. room or dispenser ID), information on how much of the dispensed substance was used (e.g. dosage size, number of towels etc.), information on who was using it (if the individual operator is tagged and sensed by the system), and/or information on what was used if it is a multi-dispenser containing, for example, both soap and paper .

The data 310 is depicted as a not completely filled circle. This should indicate that the data 310 may not be complete in the sense that not every actual usage event may be captured, since not every piece of hygiene equipment may be provided with a corresponding sensor. However, the present embodiment assumes that the captured fraction of usage data is good enough to sufficiently represent the usage of the hygiene equipment. In any way, however, also first and second usage data U, U' may be defined as described in greater detail in conjunction with the corresponding embodiments.

The system 30 of the present embodiment further comprises a second receiving section 32 which is configured to receive first opportunity data 0 from a first opportunity sensor arrangement 200 and second opportunity data 0' from a second opportunity sensor arrangement 200'. The first opportunity data indicates a first set of opportunities to use the hygiene equipment 10 and this first opportunity data 0 is collected by means of receiving individual data signals o-l, o-2, o-3,... from corresponding individual pieces of the sensor arrangement 200 such as cameras 200-1, 200-2, vicinity and/or door passing sensors 200-3, 200-4, and the like. Generally, the opportunity sensor arrangements 200, 200' may be any selection of cameras, low resolution cameras (so it may be difficult to identify individuals in the image data) , time- of-flight cameras, infrared (IR) cameras, heat/thermo- cameras, micro-phones, image recognition resources, vicinity sensors, radar, ultrasonic sensors, IR sensors, photocell sensor, conductive and/or capacitive sensors (presence, touching) , laser range sensors, a time-of-flight sensors (e.g. sensors that employ the delay of RF-, e/m-pulse or light signals for determining a location, a distance and/or movements), RFID readers and/or NFC equipment (e.g. also for identifying a badge carried by an operator) , door pass sensors, a light barrier, and the like. The collected opportunity data signals o-l, ... are then stored as first opportunity data 0 by the system 30 again in the database 34 as data 321.

Similar to the signals u-1, -2 , ... the signals o-l, o-2, o-3, are typically signaling "now there is/was an opportunity" (e.g. by carrying a Boolean value "TRUE" or by simply carrying data such as a dispenser or location ID) . In a way, the mere fact that a signal is received may indicate to the system 30 that there is an opportunity to use some piece of hygiene equipment. However, the signal may also include more information, including information on when the piece of hygiene equipment could have been used (e.g. timestamp) , information on how much of the dispensed substance should have been used (e.g. dosage size, number of towels etc.), information on who could have used it (if the individual operator is tagged and sensed by the system), and/or information on what could have been used by the operator if there are alternatives (for example, soap, towel, or disinfectant) . Alternatively or additionally, the opportunity data or signals may also include information on a physical movement of a person or object throughout the facility. In this way also indirect indications to opportunities to use the hygiene equipment may be considered.

Further, the opportunity signals 0 may also include more holistic information such as movement /location of people (identified or not identified) , door openings/passages and other such information that would enable the system to interpret the data and draw its own conclusions, based on proper training, if a hand hygiene opportunity has occurred. This may be identified as a more holistic way of looking at opportunity sensors where the corresponding embodiments may employ artificial intelligence (AI) for interpreting for example "raw" movement data. Specifically, the artificial intelligence (function, neural network, statistical learning implementation) may be given access to the raw data with as little intermediate interpretation as possible.

Likewise, opportunity data (especially the first opportunity data 0) may also be collected by one or more human observers 201 who enter the data into an input device 200-5 (keyboard, smartphone, tablet computer, etc.) that, in turn, generates and forwards a corresponding signal o-5 to the receiving section 32, either in real-time or in a subsequent analysis step by means or correlating time-stamps. It is noted that - at least from the viewpoint of the claimed system - the signal o-5 may not be distinguishable from a signal o-l or o- 3 which originates from a sensor device. In a way, the opportunity sensor equipment 200, 200' may then comprise the input device 200-5 as a further individual sensor, although, in certain embodiments, it may be preferable to have such a comparatively labor intense solution only for the first set of opportunity sensors 200. It would then be a further advantage of the corresponding embodiment that such a labor intense solution is rendered superfluous during the later operation of the system.

The system 30 further comprises a calculation section 33 that is configured to determine a function F from the usage data U and the first opportunity data 0, retrieved from, respectively, 310 and 321 of the database 34. In one embodiment, the calculation section 33 determines as the function F a set of correlation parameters that allows for estimating a compliance metric from a given set of the usage data U and second opportunity data 0' .

A general example is shown in conjunction with the flow chart of Figure 4A that is to show an exemplary way to determine parameters of a function according to an embodiment of the present invention. In steps Sll and S12 usage data "U" and corresponding first opportunity data "0" are received either sequentially or concurrently (note that the depicted order may likewise be reversed for sequential reception or even made parallel for simultaneous real-time recording so that the first opportunity data is received prior to receiving the usage data) . Optionally and either sequentially or concurrently relative to any one of steps Sll and S12, there can be received metadata in a step S15 (note that the depicted order may likewise be reversed for sequential reception so that the receiving the usage/opportunity data, determining Cc, and, optionally, receiving the meta data may take place in any suitable order) . Since the received first opportunity data provides a good quality picture for the actual opportunities, a "complete" compliance metric Cc can be calculated in step S13. The compliance metric Cc corresponds to a best obtainable compliance where a best effort (later to be reduced) is put into obtaining the opportunity data, i.e. the so/called first opportunity data "0" . The compliance metric Cc can be therefore assumed as to be the true compliance metric:

Cc = U/0;

With the knowledge of Cc, U and 0 it is then possible to determine parameters of a function F in step S14 which reproduces a compliance metric also without knowledge of the first opportunity data 0, but rather based on the reduced opportunity dataset 0' .

Generally, the first opportunity data and the corresponding usage data (first usage data or the usage data obtained during at least an overlapping period when the first opportunity data was obtained) , will allow the calculation of some kind of "complete" compliance metric Cc by employing the arrangements 100 and 200. For example, the data 0 may indicate for a given period 1000 opportunities, whilst the data U indicates for a corresponding period 920 usage events, so that the complete compliance metric Cc could be calculated to Cc = 920 / 1000 = 0.92.

The function F can be established with the purpose to reflect the experience from this "known" relationship amongst U, 0, and Cc so as to allow an estimation of a compliance rate Ce also from data of lower quality, such as 0' and/or U' . In this case the "operational" compliance rate can be estimated by relating such less complete data by means of Ce = F(0',U) or F(0',U'), depending on whether or not also the equipment sensor arrangement is changed for the operation phase relative to the learning phase. In addition to this, the calculation section 33 can also process metadata M, M' , that is, however, explained in greater detail in conjunction with the corresponding embodiments.

For this purpose, the second receiving section 32 is further configured to receive second opportunity data 0' from a second opportunity sensor arrangement 200'. This second opportunity data 0' indicates a second set of opportunities to use the hygiene equipment 10, wherein the second set of opportunities is captured with a reduced sensor arrangement relative to the first set of opportunity sensors. In other words, the second opportunity sensor arrangement 200' may be reduced as compared to the first sensor arrangement 200. For example, the sensor equipment 200 may comprise not only door passing/vicinity sensors 200-3, 200-4, and the like, but may also comprise more sophisticated pieces of sensor equipment, such as cameras 200-1, 200-2. In this way, a relatively more complete first opportunity data 0 (321 in 34) can be obtained as compared to the relatively less complete second opportunity data 0' (322 in 34) .

However, the more complex first opportunity sensor arrangement 200 may not be desirable to be operated and installed for a longer period of time, since it may be too expensive and/or time consuming to maintain or operate, require legal or ethical approvals or may also be seen as intrusive or not respectful of the personal integrity and therefore require the consent of individuals when, for example, optical and or visual surveillance equipment is used (e.g. cameras 200-1, 200-2) . Whilst the latter may be tolerable for a limited period, embodiments of the present invention allow for the general operation of the system 30 without the complex set of opportunity sensor arrangement 200, and the system 30 can be also advantageously operated by a reduced set of opportunity sensor arrangement 200', which, in turn, produces said reduced second opportunity data 0' . Embodiments of the present invention can therefore provide the advantage that the system only needs to be "trained" for a limited amount of time with relatively complex opportunity sensor arrangements, whilst it can be then operated with only relatively simple opportunity sensor arrangements. The embodiments with regard to a "learning phase" and an "operation phase" are explained in greater detail in conjunction with Figures 4A, 4B, and 4C.

In any way, however, the calculation section 33 can be configured, for example, to determine the function F as a product of a machine learning procedure. In such procedures some initial parameters are preset or randomly chosen and the resulting output (here for example the output compliance metric estimate) is compared to a target value that the function F should reproduce for a given set of input data. This is fed back to the procedure that iteratively adjusts the employed parameters so as to match the target output. In a way, during such a machine learning procedure the calculation section 33 can be "trained" so as to determine the suitable parameters.

Machine learning can be generally identified as a set of algorithms and procedures enabling a computing apparatus (computer) to make analysis and predictions based on incomplete data. Many of these algorithms, such as Artificial Neural Networks, are indeed inspired and try to mimic the function of our nerves and brain. A synonym used often alongside machine learning is the so-called "statistical learning" which is a collection of relevant base techniques that are - as such - known and documented e.g. in T. Hastie et al . : "The Elements of Statistical Learning" (2^nd edition, Springer, ISBN: 978-0-387-84857-0) .

As an example, the calculation section 33 can be thus configured to determine the function F as a set of correlation parameters from usage data U acquired for the actual use of the hygiene equipment 10 by the sensor arrangement 100 and the first opportunity data 0 measured/detected by the first opportunity sensor arrangement 200 at the same time or during at least in part overlapping intervals. In this way, the first opportunity sensor arrangement 200 may at least to some extent provide a sufficient measurement of the opportunities in order to estimate a compliance metric. At least parts of the usage data U and the first opportunity data are correlated by these parameters to the compliance, so that also further usage data and second (reduced) opportunity data acquired later can be correlated to the compliance by using the parameters of the function .

According to a further embodiment, the equipment sensor arrangement 100-1,... may comprise a first set of equipment sensors and a second set of equipment sensors, and the first receiving section 31 may be correspondingly configured to receive as said usage data first usage data from the first set of equipment sensors and second usage data from the second set of equipment sensors. The calculation section 33 may then also be configured to determine said function based on the first usage data and said first opportunity data, and to estimate said compliance metric from the second usage data and said second opportunity data. In other words, embodiments of the present invention also consider that the equipment sensor arrangement is not necessarily the same for acquiring usage data for determining the function, e.g. during a learning phase, and, respectively, for operating the system for estimating the compliance rate.

As a consequence the usage data U may also be referred to as first usage data and the database 34 further stores second usage data U' . These embodiments may specifically relate to situations where after the learning phase the number of hygiene equipment or corresponding sensors is reduced, since it may have been determined that it may be enough to have for example only a certain (smaller) number of hygiene equipment (dispensers) equipped with sensors for the purpose of estimating the compliance metric. The equipment sensor arrangement may be considered as being too expensive for later operation and one may want to cut down also on the equipment sensors apart from the opportunity data sensors. As an example, the mentioned second set of equipment sensors may be a part of said first set of equipment sensors, where some sensors of the installed hygiene equipment are simply deactivated (or no longer monitored) , or a part of the hygiene equipment is replaced after the learning phase by equipment without sensors and/or connectivity.

Figure 2 shows a schematic view of an exemplary deployment of a system for estimating a compliance metric according to another embodiment of the present invention. As an example, there is shown as a facility an intensive care unit 400 with corresponding intensive care points: first and second patient stations 410,420 and first and second patient care equipment 411,421. As can be seen, the intensive care unit 400 may be occupied by one or two patients in the shown configuration, whilst the embodiments of the present invention may naturally envisage also other intensive care units with any number of patients and personnel and/or other facilities as mentioned elsewhere in the present disclosure.

In general, however, the number of actually occupied intensive care points (beds) as well as other information (e.g. on a number of individuals working/visiting, on a shift, a time of day, a day of week, a holiday) can be considered as optional metadata M, likewise stored, for example, in the database of Figure 1. Further examples include information on a type of care given (intensive, orthopedic, surgery, child, emergency, ear/nose/throat, etc.), size of ward / floor / unit in terms of number of beds and/ number of staff, time and date, number of staff working the specific time also divided by staff category (cleaners, nurses, doctors etc.), outdoor weather, cleaning records, and data measured by corresponding equipment (e.g. a pressure/weight sensor in a bed to indicate if there is a patient in the bed or not) . Said metadata M can thus contribute in the calculation section 33 when determining the function F and/or contribute when estimating the compliance metric based on the function F and the current usage and second opportunity data U, 0' . For example, the metadata M can contribute in rendering the function F sensitive to the actual environment (e.g. beds occupied or not) so that it can produce an improved estimate Ce for situations when all or most beds are occupied and when only some beds are occupied.

In an embodiment, the metadata M contributes to the accuracy of the function F in a way that specific scenarios of metadata M/M' and opportunity data 0/0' and/or user data U/U' and correspond to respective opportunity scenarios. For example, the metadata M/M' is obtained through receiving signals from pressure (or heat) sensors 412, 422. If one or more of such sensor signals indicate that a corresponding patient station 410, 420 is in use, the opportunities for the use of hygiene equipment will accordingly change. For example, an algorithm may correlate signals from the vicinity sensor 202 and the sensors 412, 422, in a way that signals that indicate an activation of the vicinity sensor 202 (i.e. operator present) and a first pressure sensor 412 correlate to opportunities only in connection with the first patient station 410, whilst signals that indicate an activation of the vicinity sensor 202 and the first pressure sensor 412 and the second pressure sensor 422 correlate to opportunities in connection with both patient stations 410, 420. Accordingly, the parameters of the function F may be compiled in the way that the metadata influences the opportunity scenarios. This is schematically depicted in Figure 4A, where an optional step S15 includes receiving of the metadata during learning, e.g. when the function and/or its corresponding parameters is/are determined. The configuration shown in Figure 2 may be representative for a first learning phase during which the system acquires usage data from an equipment sensor arrangement provided for one or more of the individual pieces of hygiene equipment, such as the soap dispenser 102, the first and second disinfectant dispensers 103, 104, and the hand washing sink 101. In this way, the system is able to receive usage data U from these pieces of equipment 101 - 104 as possibly individual signals from each corresponding device/sensor. During this phase, also a first opportunity sensor arrangement is provided in the intensive care unit 400 that comprises one or more cameras 201, 203, a vicinity sensor 202, and a door passing sensor 204. Thereby, the first camera 201 may be in particular arranged for detecting an opportunity in a dedicated area, such as the surrounding area 2001 of the hand wash sink 101. For example, the image and/or video data obtained from first camera 201 may be processed or analyzed for determining whether an individual could/ should have used the soap dispenser 102 when washing his/her hands at sink 101. In a way, the use of the sink 101 implies also an opportunity to use the soap dispenser 102. However, the configuration as shown in Figure 2 is only to be seen as an application example and the embodiments of the present inventions may well be applicable also to configurations different from the environment of an intensive care unit in particular, or from a hospital in general.

Similarly, second camera 203 may observe a further dedicated area 2002 that covers the vicinity and area of second patient station 420, which can be, for example, a bed. In an example, the corresponding image and/or video data from the second camera 203 may be processed or analyzed in order to find an individual entering the area of the second patient station 420 and/or determining the duration and time for how long the individual remains in the vicinity thereof. This could likewise imply an opportunity to use the first disinfectant solution dispenser 104 before or at the early stage of entering the area 2002 that covers the vicinity and area of the second patient station 420. Such visual determination of an opportunity by images and/or a human observer may also be selected for opportunities that are difficult to detect by sensors, such as a healthcare worker performing an aseptic task (WHO moment Nr. 2) or after a body fluid exposure risk (WHO moment Nr. 3) . It may be preferable to further consider that WHO moments Nr . s 1 & 5 may be correlated to moments 2 & 3, which can be exploited for removing sensors that either sense moments 1 & 5 or 2 & 3 in the second opportunity sensor arrangement. Such correlations may also be made automatically by the machine learning algorithm.

Likewise, a vicinity sensor 202 may determine the opportunity to use the second disinfectant dispenser 104 when an individual operates first patient care equipment 411 which, in turn, can indicate that manual operations or actions are carried out to patient in first patient station 410, which can be, for example, a bed. In general, any one of the equipment sensors and opportunity sensors may convex signals in any suitable manner, such as by wire-bound communication or wireless communication as, for example, shown between the first disinfectant dispenser 103 and a wireless data acquisition and collection point 105.

As already mentioned, however, the use of first and second cameras 201, 203, or in general a complex opportunity sensor arrangement, may be problematic for various reasons. For example, the analysis and processing of the corresponding and produced image or video data may be expensive, since, for example, sufficiently powerful image processing hardware needs to be employed or a human operator may need to view the image data so as to "manually" determine the corresponding opportunity data. Furthermore, the use of such cameras may require the consent of individuals and personnel being present or active in the intensive care unit 400. At the same time, however, it may be tolerable that cameras are installed for a limited time so that the more expensive opportunity sensor arrangement can be deployed for an initial learning phase, when a calculation section of the system determines a function F based on usage data received from the hygiene equipment 101,... and the opportunity data (first opportunity data) from the first opportunity sensor arrangement comprising also at least the first and second cameras 201, 203.

At a later (operation) stage, the deployed opportunity sensor arrangement may be reduced in the sense that some individual pieces of opportunity sensor equipment is removed and/or deactivated. For example, at a second phase, the first and second cameras 201 and 203 may be unmounted so that a specific consent to individuals and or personnel in the context of video surveillance may no longer be necessary. Likewise, any expensive or burdensome analysis of the corresponding image or video data may no longer be necessary. At the same time, however, a somewhat reduced second opportunity sensor arrangement remains active within the intensive care unit 400 as, for example, consisting only of the vicinity sensor 202 and the door passing sensor 204. In this embodiment, the second opportunity sensor arrangement is a direct reduction from the first opportunity sensor arrangement, or, in other words, the second opportunity sensor arrangement is a part of the first opportunity sensor arrangement. Furthermore, metadata sensors may be still employed, such as a pressure/weight sensor for determining whether a bed is occupied or not.

An embodiment of the present invention now envisages to estimate a compliance metric from the usage data that may well be still available to a high degree of completeness and a somewhat reduced (second) opportunity data received only from the vicinity sensor 202 and door passing sensor 204 from the function F that was previously determined based on the usage data and the more comprehensive (first) opportunity data when the first and second cameras 201 and 203 were still operative .

Figures 3A and 3B show schematic views in conjunction with a neural network being used for determining a function according to a further embodiment of the present invention. In a way, the employment of a neural network is one way of determining the function as a product of a machine learning procedure. Specifically, Figure 3A schematically shows a node (neuron) 331 of a neural network. As is as such known, the neuron 331 has one or more inputs 332 and one output 333. In general, the neuron 331 receives input values CC₁₃ at the corresponding input 332, multiplies each input value CC₁₃ by a corresponding coefficient w₁₃ and forms the corresponding sum β₃ = ∑i w_±J x _±J at the output 333. Furthermore, the output β₃ could also be normalized to a value between 0 and 1, or be made binary so that it either assumes 0 or 1 (β₃ e {0, 1}) by means of applying a rounding and/or Heavyside function. As then shown in Figure 3B, a neural network 334 is composed of a corresponding manifold of neurons 331 as one is individually shown in Figure 3A. As a consequence, the network 334 provides at the bottom a number (e.g. k) of inputs 332' receiving the ¾i (at k inputs) and, after one or more hierarchy levels of individual neurons 331, at the output 333' the output βι of the topmost neuron of the network 334.

According to an embodiment of the present invention, the calculation section of the system employs such a neural network 334 for both determining the function F from initial

(learning) data 335 comprising user data U, first opportunity data 0, and, optionally, first meta data M. Since in this phase both the user data U and the first opportunity data 0 can be directly correlated to a specific compliance metric Cc

(e.g. a use rate derived from U divided by an opportunity rate derived from 0) , the neural network 334 can be trained with this data to determine the coefficient w±_j. In other words, the w_j are trained so that the network 334 gives the correct Cc at output 333' for the given U, 0, and, optionally, M. It is noted that in the context of the present embodiment, a group of or the entirety of the determined coefficients w_j corresponds to the function F as mentioned elsewhere in the present disclosure. In other words, the neural network 334 with the trained coefficients w_j represents the function F in this embodiment.

The present embodiment then envisages to employ the trained neural network 334 (function) to determine at the output 333' a value Ce of an estimated compliance metric based on a second set of data 336 that comprises (only) the second opportunity data 0' , corresponding user data U, and, optionally, again corresponding meta data M' . It is noted that the usage data U as part of data 335 may be recorded at the same time, or at least at overlapping intervals, during which the first opportunity data 0 and the meta data M is obtained. Likewise, the user data of the data set 336 may be determined (measured, detected) while the second opportunity data 0' , and, optionally, the second meta data M' is obtained. In a certain embodiment, the time periods during which 0 and U are measured are the same.

In an operation phase, the function F can then be employed to provide the compliance metric Ce from 0' , U' (and optionally M' ) so as to be an accurate estimate to a corresponding "true" or "complete" compliance metric Cc. An implementation would consider to feed the function with 0' , U' (and optionally M' ) as inputs to the function F in the same way as in the training phase (see above) . However, since the function F is now trained and the corresponding parameters have been learned, an algorithmic implementation would need to evaluate the mentioned inputs with the learned parameters. The latter learned parameters provide that the output compliance metric Ce is calculated taking into account the learned correlations and that the output is an accurate mapping of the compliance metric Cc even without the availability of the first opportunity data 0. In other words, Figures 3A and 3B depict two process steps in one, namely the training step which calculates the F in an iterative process between the "complete" initial set of learning data 335 in relation to a known Cc (at the top of the triangle) , and an operational step which uses the known F (in the triangle) which calculates (or estimates) an estimated Ce with the aid of the "incomplete" set of data 336.

With now reference to Figure 4B it is described a flowchart of an exemplary way to employ the learned function according to an embodiment of the present invention. In steps S21 and S22 usage data "U' " and corresponding second opportunity data "0' " are received either sequentially or concurrently (note that the depicted order may likewise be reversed for sequential reception so that the second opportunity data is received prior to receiving the usage data) . It is to be further noted, that the receiving of metadata in the optional step S25 and its corresponding consideration may only make sense when the function was learned with taking into account metadata, as, for example, described in conjunction with Figure 4A and the optional step S15. Since this embodiment considers that the function F (or its corresponding parameters) is already trained, it can now be referred to this function in step S23, and the compliance metric Ce can be calculated in step S24 by, for example, computationally evaluating the function:

Ce = F (IT , 0' [, M' ] ) , where [, M' ] means that the consideration of metadata M' is optional .

Figure 4C shows a flowchart of a general method embodiment of the present invention. The method embodiments can be implemented for estimating a compliance metric indicating the usage of hygiene equipment by one or more operators. For this purpose, the method embodiments comprise a first set of steps: a step Sill of receiving usage data from an equipment sensor arrangement, said usage data indicating a usage of said hygiene equipment; a step S121 of receiving first opportunity data from a first opportunity sensor arrangement, said first opportunity data indicating a first set of opportunities to use said hygiene equipment; and a step S130 of determining, based on said usage data received in the step Sill and said first opportunity data, a function for estimating said compliance metric from usage data and second opportunity data. These steps may be identified as being part of a "learning phase" during which a relatively complex opportunity arrangement is available and during which the function is determined.

The determined function can be then employed during an "operation phase". Such a phase may be associated with a second set of steps: a step S112 of receiving usage data from the equipment sensor arrangement and a step S122 of receiving second opportunity data from a second opportunity sensor arrangement, said second opportunity data indicating a second set of opportunities to use said hygiene equipment, wherein the second opportunity sensor arrangement is reduced relative to said first opportunity sensor arrangement. In a step S140 it is then estimated the compliance metric from usage data received in the second step S112 and said second opportunity data using said function. Additional phases may be envisaged, where the compliance metric estimated by using the function is verified or confirmed, and/or where the function is optimized and/or fine-tuned for the respective purpose.

According to a further embodiment of the present invention, the system is trained with the aim of reaching a stable and robust function F. In an additional test phase it can be determined how reliable the F-function is. This is done by exposing the F-function to another, un-seen, set of "complete" data (0, U) to measure how well the estimated compliance Ce matches the "real" compliance Cc. Then, the operation phase can be initiated as described elsewhere in the present disclosure. In this embodiment the "complete" data can be split in two sub-sets of which one is used in the first training/learning phase and the other in the testing phase .

Figure 5 shows a schematic view of a general entity embodiment of the present invention. The entity can be any collection of processing and memory resources that are suitable for implementing the corresponding sections of a system for estimating the compliance metric. For example, the entity 30 can be implemented as a stand-alone computer, a server, a processing share of a datacenter, or an application running on some kind of shared hardware. More specifically, the entity 30 according to the present embodiment comprises processing resources 301 (e.g. CPU), memory resources 302 and communication means 303 (e.g. a receiver/transmitter working according to WLAN, WiFi, WiMAX, Bluetooth™, GPRS, GSM, PCS, DECT, UMTS, 3G/4G/5G, LTE, etc., or a wire-bound standards such as Ethernet and the like) that are configured to communicate with some kind of network 304 (e.g. LAN, wireless communication system, an intranet, the Internet, and the like) . By means of the latter, the system is able to receive the usage and opportunity data signals u-i, o-i, etc., access the database 34, or to convey any estimated metric to a given location .

Specifically, the memory resources 302 are adapted to store code that instructs the processing resources 301 during operation to implement at least a first receiving section configured to receive usage data from an equipment sensor arrangement, said usage data indicating a usage of said hygiene equipment; a second receiving section configured to receive first opportunity data from a first opportunity sensor arrangement, said first opportunity data indicating a first set of opportunities to use said hygiene equipment, and configured to receive second opportunity data from a second opportunity sensor arrangement, said second opportunity data indicating a second set of opportunities to use said hygiene equipment, wherein the second opportunity sensor arrangement is reduced relative to said first opportunity sensor arrangement; and a calculation section configured to determine, based on said usage data and said first opportunity data, a function for estimating said compliance metric from said usage data and said second opportunity data. Naturally, the code may be adapted so as to implement any other modification envisaged by the embodiments of the present invention.

Although detailed embodiments have been described, these only serve to provide a better understanding of the invention defined by the independent claims and are not to be seen as limiting .

Claims

1. A system for estimating a compliance metric indicating the usage of hygiene equipment by one or more operators, the system comprising:

- a first receiving section configured to receive usage data from an equipment sensor arrangement, said usage data indicating a usage of said hygiene equipment;

- a second receiving section configured to receive first opportunity data from a first opportunity sensor arrangement, said first opportunity data indicating a first set of opportunities to use said hygiene equipment, and configured to receive second opportunity data from a second opportunity sensor arrangement, said second opportunity data indicating a second set of opportunities to use said hygiene equipment, wherein the second opportunity sensor arrangement is reduced relative to said first opportunity sensor arrangement; and a calculation section configured to determine, based on said usage data and said first opportunity data, a function for estimating said compliance metric from said usage data and said second opportunity data.

2. The system according to claim 1, wherein said calculation section is further configured to estimate said compliance metric from said usage data and said second opportunity data .

3. The system according to claim 1 or 2, wherein the first and second opportunity sensor arrangements comprise any selection of a camera, image recognition resources, a vicinity sensor, a door pass sensor, a light barrier, a time-of-flight sensor, and RFID readers and/or NFC equipment .

4. The system according to any one of claims 1 to 3, adapted to receive said first and said second set of opportunity data wherein said second opportunity sensor arrangement is a subset of said first opportunity sensor arrangement.

5. The system according to claim 4, wherein the first opportunity sensor arrangement comprises any selection of a camera, a video camera, a heat camera, image recognition resources, a vicinity sensor, a door pass sensor, a light barrier, and RFID readers and/or NFC equipment, and wherein the second opportunity sensor arrangement comprises any selection of a vicinity sensor, heat sensitive (IR) camera, a door pass sensor, a light barrier, a time-of-flight sensor, and RFID tags and antennas/readers and/or NFC equipment.

6. The system according to any one of claims 1 to 5, wherein the equipment sensor arrangement comprises a first set of equipment sensors and a second set of equipment sensors, and wherein the first receiving section is configured to receive as said usage data first usage data from the first set of equipment sensors and second usage data from the second set of equipment sensors.

7. The system according to claim 6, wherein said calculation section is configured to determine said function based on the first usage data and said first opportunity data, and to estimate said compliance metric from the second usage data and said second opportunity data.

8. The system according to any one of claims 1 to 7, wherein the hygiene equipment comprises any one of a soap dispenser, a dispenser for disinfectant solutions, gels or substances, a towel dispenser, a glove dispenser, a tissue dispenser, a hand dryer, a sink, a tap, a UV disinfector, and a radiation assisted disinfectant point.

9. The system according to any one of claims 1 to 8, wherein the calculation section is configured to determine said function as a set of correlation parameters.

10. The system according to any one of claims 1 to 8, wherein the calculation section is configured to determine said function as a product of a machine learning procedure.

11. The system according to claim 10, wherein the calculation section is configured to determine said function as a set of coefficients of a neural network.

12. The system according to claim 11, wherein the calculation section is configured to estimate said compliance metric from said usage data and said second opportunity data using the set of coefficients.

13. The system according to claim 10, wherein the calculation section is configured to estimate said compliance metric from said usage data and said second opportunity data using a set of coefficients of said machine learning procedure .

14. The system according to any one of claims 1 to 13, wherein the calculation section is further configured to process metadata when determining the function and/or when estimating said compliance metric.

15. The system according to any one of claims 1 to 14, wherein the usage data includes information on when the piece of hygiene equipment was used, preferably in the form of a timestamp, and/or information on where the piece of hygiene equipment was used, preferably in the form of a room or dispenser identifier.

16. The system according to any one of claims 1 to 15, wherein the system further comprises a database arranged to store usage data, opportunity data, and/or metadata.

17. A method for estimating a compliance metric indicating the usage of hygiene equipment by one or more operators, the method comprising: a first step of receiving usage data from an equipment sensor arrangement, said usage data indicating a usage of said hygiene equipment;

- a step of receiving first opportunity data from a first opportunity sensor arrangement, said first opportunity data indicating a first set of opportunities to use said hygiene equipment; a step of determining, based on said usage data received in the first step and said first opportunity data, a function for estimating said compliance metric from usage data and second opportunity data; a second step of receiving usage data from the equipment sensor arrangement; a step of a receiving said second opportunity data from a second opportunity sensor arrangement, said second opportunity data indicating a second set of opportunities to use said hygiene equipment, wherein the second opportunity sensor arrangement is reduced relative to said first opportunity sensor arrangement; and a step of a estimating said compliance metric from usage data received in the second step (S112) and said second opportunity data using said function.