Exit monitoring system for patient support apparatus
An exit prediction system receives movement data, divides the movement data into segments of time, extracts features from each segment of time, and determines a pattern of movement from the extracted features. A patient exit from a patient support apparatus is predicted based on the determined pattern of movement.
1. An exit prediction system, comprising:
a bed including:
a mattress;
load cells positioned under the mattress of the bed; and
a control system included on the bed, the control system including:
at least one processor; and
a memory storing instructions which, when executed by the at least one processor, cause the system to:
receive movement data from the load cells;
divide the movement data into segments of time;
extract features from each segment of the segments of time;
determine a pattern of movement from the extracted features;
generate a patient exit score based on the determined pattern of movement changing according to a predetermined exit pattern; and
trigger an alert when the patient exit score exceeds a threshold, wherein the alert is generated on the bed based on the patient exit score.
2. The system of claim 1 , wherein the memory stores further instructions which, when executed by the at least one processor, cause the system to:
generate low, moderate, or high alerts based on the patient exit score.
3. The system of claim 2 , wherein the low, moderate, or high alerts are customizable based on factors related to patient fall and injury likelihood when exiting the bed unattended.
4. The system of claim 1 , wherein the pattern of movement is configured to be evaluated by a bed exit prediction model built by machine learning, deep learning, or artificial intelligence.
5. The system of claim 4 , wherein the bed exit prediction model is built using motion profiles that are pre-categorized as non-exit motions or exit motions.
6. The system of claim 5 , wherein the bed exit prediction model is calibrated based on multiple speed classifiers for various patient speeds.
7. The system of claim 1 , wherein the extracted features from each segment of time include a percentage change in load detected by the load cells in the bed.
8. The system of claim 7 , wherein the memory stores further instructions which, when executed by the at least one processor, cause the system to:
assign output classes to each segment of the segments of time, each output class of the output classes indicating a decision corresponding to an exit motion type or a non-exit motion type.
9. The system of claim 8 , wherein the memory stores further instructions which, when executed by the at least one processor, cause the system to:
classify the pattern of movement as a non-exit pattern of movement or an exit pattern of movement based on the output classes over time.
10. The system of claim 1 , wherein the system is part of a centralized caregiver call system that generates alerts based on the patient exit score.
11. A method of predicting a patient exit from a bed, the method comprising:
receiving motion profiles, the motion profiles being detected by a plurality of load cells positioned under a mattress of the bed;
pre-categorizing the motion profiles as non-exit motion or exit motion;
extracting features from the pre-categorized motion profiles;
generating a bed exit prediction model using the extracted features;
predicting the patient exit from the bed using the model by generating a patient exit score based on a pattern of movement changing according to an exit pattern identified by the model; and
triggering an alert when the patient exit score exceeds a threshold, wherein the alert is generated on the bed based on the patient exit score.
12. The method of claim 11 , further comprising:
generating low, moderate, or high alerts based on the patient exit score.
13. The method of claim 12 , further comprising:
customizing the low, moderate, or high alerts based on additional factors related to patient fall and injury likelihood when a patient exits the bed unattended.
14. The method of claim 11 , wherein the bed exit prediction model identifies the exit pattern from the extracted features using machine learning, deep learning, or artificial intelligence.
15. The method of claim 14 , wherein the exit pattern is identified by dividing the pre-categorized motion profiles into segments of time, and analyzing a row of features for each segment of the segments of time.
16. The method of claim 15 , wherein predicting the patient exit from the bed includes assigning output classes to the segments of time within a motion profile detected from the bed, each output class of the output classes indicating a decision corresponding to an exit motion type or a non-exit motion type.
17. A non-transitory computer readable storage media, including computer readable instructions which when read and executed by a computing device, cause the computing device to:
receive motion profiles, the motion profiles being detected by a plurality of load cells positioned under a mattress of a bed;
pre-categorize the motion profiles as non-exit motion or exit motion;
extract features from the pre-categorized motion profiles;
generate an exit prediction model using the extracted features; and
predict a patient exit from the bed using the exit prediction model by generating a patient exit score based on a pattern of movement changing according to an exit pattern identified by the exit prediction model; and
trigger an alert when the patient exit score exceeds a threshold, wherein the alert is generated on the bed based on the patient exit score.
18. The non-transitory computer readable storage media of claim 17 , further comprising computer readable instructions which when read and executed by the computing device, cause the computing device to:
generate low, moderate, or high alerts based on the patient exit score.
19. The non-transitory computer readable storage media of claim 18 , further comprising computer readable instructions which when read and executed by the computing device, cause the computing device to:
customize the low, moderate, or high alerts based on additional factors related to patient fall and injury likelihood when a patient exits the bed unattended.
20. The non-transitory computer readable storage media of claim 17 , wherein to predict the patient exit from the bed includes to assign output classes to segments of time within a motion profile detected from the bed, with each output class of the output classes indicating a decision corresponding to an exit motion type or a non-exit motion type.