Method and system for computer-aided triage of stroke
A system for computer-aided triage includes a router, a remote computing system, and a client application. A method for computer-aided triage includes receiving a data packet associated with a patient and taken at a point of care; checking for a suspected condition associated with the data packet; in an event that the suspected condition is detected, determining a recipient based on the suspected condition; and transmitting information to a device associated with the recipient.
1. A method for the automated triaging of a patient based on a vessel occlusion type, the method comprising:
retrieving a data packet associated with the patient from a data collection device;
processing the data packet with a set of multiple trained machine learning algorithms comprising a first and a second trained machine learning algorithm, the second trained machine learning algorithm separate and distinct from the first;
detecting a vessel occlusion and categorizing the vessel occlusion type based on outputs of the set of multiple trained machine learning algorithms, wherein the outputs comprise at least a location of the vessel occlusion relative to a set of anatomical landmarks;
based on the vessel occlusion type, selecting an automated alert from a set of multiple automated alerts;
transmitting the automated alert to a set of user devices associated with a set of users; and
prompting treatment of the patient based on at least one input received from the set of user devices in response to the automated alert.
2. The method of claim 1 , wherein the second trained machine learning algorithm receives additional inputs relative to the first trained machine learning algorithm.
3. The method of claim 2 , wherein the first trained machine learning algorithm is configured to detect a large vessel occlusion, wherein the large vessel occlusion comprises an occlusion at an M 1 segment of a middle cerebral artery.
4. The method of claim 2 , wherein each of the first and second trained machine learning algorithms comprises a convolutional neural network.
5. The method of claim 4 , wherein:
the first trained machine algorithm receives a set of images of a brain of the patient; and
the second trained machine learning algorithm receives the set of images and a mirror image version of the set of images.
6. The method of claim 1 , wherein:
the first trained machine algorithm receives a set of images of a brain of the patient; and
the second trained machine learning algorithm receives the set of images and a processed version of the set of images.
7. The method of claim 1 , further comprising calculating a severity score associated with the patient based on the outputs, wherein selecting the automated alert is further performed based on the severity score.
8. The method of claim 7 , further comprising calculating an irreversibly damaged volume of brain with the second trained machine learning algorithm, wherein the severity score is further calculated based on the irreversibly damaged volume.
9. The method of claim 7 , further comprising adjusting a place of the patient in a queue of multiple patients based on the severity score, wherein at least one of a timing of the automated alert or a subset of devices in the set of devices is adjusted in response to the adjusted place of the patient.
10. The method of claim 1 , wherein the data packet is retrieved at a processor operable as a virtual machine.
11. The method of claim 1 , wherein the data collection device is located at a first point of care and wherein treatment of the patient is performed at a second point of care separate and distinct from the first point of care.
12. A system for the automated triaging of a patient based on a vessel occlusion severity, the system comprising:
a set of multiple trained machine learning algorithms;
a processing subsystem operable, at least in part, as a virtual machine, wherein the processing subsystem:
receives a first data packet associated with the patient from a data collection device;
produces a second data packet with the first data packet;
retrieves a trained machine learning algorithm from the set;
evaluates the trained machine learning algorithm with the first and second data packets;
detects a vessel occlusion and categorizes the vessel occlusion severity based on outputs of the trained machine learning algorithm;
based on the vessel occlusion severity, selects an automated alert from a set of multiple automated alerts;
transmits the automated alert to a set of user devices associated with a set of users; and
prompts treatment of the patient based on at least one input received from the set of user devices in response to the automated alert.
13. The system of claim 12 , wherein the outputs comprise a calculated irreversibly damaged volume of brain, wherein the vessel occlusion severity is determined at least in part based on the irreversibly damaged volume.
14. The system of claim 13 , wherein the outputs further comprise a location of the vessel occlusion relative to a set of anatomical landmarks.
15. The system of claim 12 , wherein the treatment comprises initiating a transfer of the patient to a second point of care, wherein the first data packet is collected at a first point of care remote from the second point of care.
16. The system of claim 12 , wherein the second data packet is a mirror image version of the first data packet.
17. The system of claim 12 , further comprising processing the data packet with a second trained machine learning algorithm of the set, wherein the second trained machine learning algorithm is evaluated with the first data packet and is not evaluated with the second data packet.
18. The system of claim 17 , wherein the second trained machine learning algorithm is configured to detect a large vessel occlusion, wherein the large vessel occlusion comprises an occlusion at an M 1 segment of a middle cerebral artery.
19. The system of claim 17 , wherein each of the first and second trained machine learning algorithms comprises a convolutional neural network.
20. The system of claim 12 , wherein the processing subsystem further adjusts a place of the patient in a queue of multiple patients based on the vessel occlusion severity, wherein at least one of a timing of the automated alert or a subset of devices in the set of devices is adjusted in response to the adjusted place of the patient.