5.1 Stasis monitoring
By measuring the center of gravity of the bed, the system can determine if the patient remains motionless for extended periods of time (stasis), posing health hazards. The control center will determine if the situation is normal or whether immediate action is required.
5.2 Tremor monitoring
In the event that the patient suffers from abnormal tremors, such as those caused by the onset of high fevers or seizures (for patients with this type of pathology), the system, which is equipped with a special sensor for monitoring bed vibrations, sends an alert to the central requiring to take the necessary measures.
5.3 Weight tracking
The technology may also keep track of the patient's weight. This feature enables the structure to assess body mass loss and weight gain owing to factors such as water retention. This information enables medical personnel to better assess the patient's status and consequently take relevant steps. It should be emphasized that the system is capable of distinguishing whether the weight fluctuation is actual or due to the sudden introduction of alien bodies such as books, mobile phones, and so on, or exogenous variables such as drinking a glass of water.
5.4 Standardization
The system has been developed to work with beds from any manufacturers. In this respect, the load cell is integrated in a tiny platform (3D printed in APS-PRO) with safety edges, on which the bed's braked wheel (or the bed's foot) is mounted. Another three equal-thickness passive platforms are utilized to balance the remaining feet of the bed.
5.5 Quality of sleep monitoring
The System has recently been improved with the ability to monitor the quality of sleep, in order for doctors to assess the possibility of administering to the patients appropriate medications to provide them with a good night's rest.
5.6 Impatience / psychomotor agitation monitoring
Another function of primary importance with which the system has been equipped is the ability to connect the effects of specific pharmaceuticals with any states of impatience / psychomotor agitation in the patient, allowing clinicians to better define the dosage or employ replacement treatments that are more tolerated.
5.7 Software features
A specific application was developed to allow central officials to maintain the beds under continual supervision by monitoring the patients various states in real time. Figure 4 depicts the screen for two rooms on a particular floor. The color coding denotes the various states of the bed-patient system. It is augmented in urgent situations by particular sound signals or even sirens.
Because of the IOT adoption, it is also feasible to integrate any departments and facilities, even those located at a vast distance. The system is simple and intuitive for the operators, and the notifications may be tailored to their own preferences.
5.8 Platform configurator
The user interface shown in Fig. 4 is intended to be adjusted by the surveillance Operator using a specific configurator (Fig. 5), which allows the map to be readily set up. To speed up procedures while maintaining complete granularity, settings may be created for single beds or for groups by easily specifying the number of beds that the Operator wants to act similarly.
5.9 Digital Twins
The team has decided to exploit the data collected from beds (online and real time) to develop the Digital Twin of each bed, thus introducing another of the most important 4.0 features.
The objectives of this choice are multiple:
- Conduct an experimental campaign extended in the number of beds and over time
- Anticipate digital changes on the system, faster and less expensive than physical ones
- Extend the “bed system” to a system of multiple integrated beds, so creating a simulator that includes the DT (Digital Twin) of each bed, thus implementing a CPS (Cyber Physical System)
- Create a system capable of learning from the development of daily events, to improve the safety and health of the bedridden patient (by enhancing the DT, and therefore the CPS, with algorithms of Artificial Intelligence, Machine Learning and Deep Learning)
5.10 Cyber Physical System
The purpose of the CPS is to continuously compare the real environment, made up of beds and patients, with the virtual environment, reconstructed in the Cyber Space (emulated on a server). CPS widely facilitates decision making on patients’ management, minimizing, where not needed, the physical presence of Operators. Another advantage provided by the CPS consists in the centralization and integration of data to develop an overall analysis that transforms the fragmented data into usable and actionable information. Such information is then consolidated into knowledge, with the dual purpose of being able to locally control the feedback process (for example by activating notifications or sirens, in an emergency) or to make remote structures interconnectable and comparable, thus opening up huge benchmarking opportunities for standardization, learning, improvement of customer service and cost reduction.
5.11 Artificial intelligence
The first implementation step following the development of the DTs consisted in improving the system with cognitive algorithms, capable of further limiting the need for surveillance of the beds, conducted in the presence of operators. In fact, these algorithms replace the Operator in decision-making (e.g. to evaluate whether an attempt to climb over the bank can be successful, so generating an effective risk, rather than false alarmism) and demand for urgent intervention only in case of real need. For this purpose it was decided to use the platform made available by Microsoft "Project Bonsai" which made it possible to improve the overall efficiency of the assistance process, by improving the availability of the surveillance staff through the reduction of downtime.
5.12 Machine Learning
The capability to anticipate accidents by preventing any potentially dangerous actions of bedridden patients by learning about changing their habits, such as movement, agitation, average number of toilet visits, number of attempts to leave the bed or number of loads of the safety rails. For this purpose, an analytical environment dedicated to the Statistical Process Control of the patient's habits was implemented, capable of defining typical behaviors, based on the average of the parameters detected. Control Limits are defined to distinguish normal behavior (tolerable range of variation attributable to common causes) from anomalous behavior.
5.13 Deep learning
It consists in making the machine autonomous in the identification of new influencing factors that the authors may not have initially considered for any reason as, for example, the lack of data. It is about the implementation of neural networks capable of extracting (through the analysis of a vast amount of data collected from several read and compared to each other) patterns, contrary invisible, and from them deducing the impacting factors, automatically evaluating their significance and classifying them in order of priority.