Methods, systems, and computer readable media for managing medical devices

1 . A system for managing a plurality of medical devices, the system comprising:

a server including at least one processor implemented in hardware;

a stream processor implemented by the at least one processor on the server for receiving medical data from a plurality of medical devices and for generating metadata and storing the medical data and the metadata for each of the devices;

data storage implemented in hardware on the server for storing the medical data and the metadata;

a communications interface implemented by the at least one processor on the server for implementing at least one communications protocol for providing access to the medical data and the metadata; and

at least one application implemented by the at least one processor on the server for accessing the medical data and the metadata via the communications interface and generating, based on the medical data and the metadata, output indicative of operating status of the medical devices, wherein the at least one application includes an anomaly detector configured to monitor connection status of the medical devices, predict maintenance events concerning the medical devices as anomalies, prompt a user to confirm that the maintenance events occur as expected, and

responsive to receiving user input indicating that the maintenance events occur as expected, label data corresponding to the maintenance events as expected anomalies, generate a fine-grained calibration data set including the data corresponding to the maintenance events labeled as expected anomalies and data corresponding to other connection events as unexpected anomalies, and use the fine-grained calibration data set to adaptively recalibrate the anomaly detector; and

wherein the at least one application includes a ventilation alert application for extracting, in real time and from messages sent by ventilators, an expired tidal volume variable value, determining whether the expired tidal volume variable value is greater than zero, and, in response to determining that the expired tidal volume variable value is greater than zero, applying false positive detection logic that utilizes a patient name and a previous ventilator stop time to detect a false positive, and, when the false positive detection logic does not detect a false positive, sending an alert to a virtual intensive care unit.

2 . The system of claim 1 wherein the medical data comprises health level seven (HL7) data.

3 . The system of claim 1 the stream processor is configured to determine a device identifier and an arrival time of medical data from each of the medical devices and wherein the metadata includes a timestamp history for each of the medical devices.

4 . The system of claim 3 wherein the metadata includes a latest timestamp for each of the medical devices.

5 . The system of claim 1 wherein the stream processor is configured to store the medical data in a time series database.

6 . The system of claim 1 wherein the communications interface is configured to provide access to the medical data via a representational state transfer (REST) or a sockets application programming interface (API).

7 . The system of claim 1 wherein the at least one application comprises a medical device dashboard for graphically displaying the output indicative of the operating status of the medical devices.

8 . The system of claim 7 wherein the output indicative of the operating status includes an online status and a time when a last message was received from each of the medical devices.

9 . The system of claim 7 wherein the medical devices include medical devices of different types and from different medical device manufacturers.

10 . The system of claim 7 wherein the output indicative of the operating status includes an indication of a time between receipt of messages from a medical device.

11 . The system of claim 7 wherein the output indicative of the operating status includes an indication of a number of active medical devices connected to a server.

12 . The system of claim 1 wherein the anomaly detector is configured for monitoring usage patterns of the medical devices from the medical data, for comparing the usage patterns to usage patterns reconstructed based on models, and for indicating an anomaly if a monitored usage pattern differs from any of the reconstructed usage patterns by more than a threshold amount.

13 . The system of claim 12 wherein the anomaly detector is configured to generate an anomaly score for a detected instance of an anomaly, determine whether the anomaly score satisfies a probably approximately correct (PAC) guarantee based on a false alarm rate (FAR) and a miss alarm rate (MAR), and output an anomaly prediction based on the PAC guarantee.

14 . A method for medical device management, the method comprising:

receiving, by a stream processor implemented on at least one processor of a server, medical data from a plurality of medical devices and generating metadata and storing, in data storage implemented in hardware on the server, the medical data and the metadata for each of the devices, wherein the at least one processor of the server is implemented in hardware;

providing a communications interface implemented by the at least one processor on the server for implementing at least one communications protocol for accessing the medical data and the metadata; and

providing at least one application implemented by the at least one processor on the server for accessing the medical data and the metadata via the communications interface and generating, based on the medical data and the metadata, output indicative of operating status of the medical devices wherein the at least one application includes an anomaly detector configured to monitor connection status of the medical devices, predict maintenance events concerning the medical devices as anomalies, prompt a user to confirm that the maintenance events occur as expected, and

responsive to receiving user input indicating that the maintenance events occur as expected, label data corresponding to the maintenance events as expected anomalies, generate a fine-grained calibration data set including the data corresponding to the maintenance events labeled as expected anomalies and data corresponding to other connection events as unexpected anomalies, and use the fine-grained calibration data set to adaptively recalibrate the anomaly detector;

wherein the at least one application includes a ventilation alert application, and further comprising extracting, by the ventilation alert application, in real time, and from messages sent by ventilators, an expired tidal volume variable value;

determining, by the ventilation alert application, whether the expired tidal volume variable value is greater than zero;

in response to determining that the expired tidal volume variable value is greater than zero, applying, by the ventilation alert application false positive detection logic that utilizes a patient name and a previous ventilator stop time to detect a false positive; and

when the false positive detection logic does not detect a false positive, sending, by the ventilation alert application, an alert to a virtual intensive care unit.

15 . The method of claim 14 wherein the medical data comprises health level seven (HL7) data.

16 . The method of claim 14 wherein generating the metadata includes determining a device identifier and an arrival time of medical data from each of the medical devices and wherein the metadata includes a timestamp history for each of the medical devices.

17 . The method of claim 16 wherein the metadata includes a latest timestamp for each of the medical devices.

18 . The method of claim 14 comprising storing the medical data and the metadata in a time series database.

19 . The method of claim 14 wherein providing the communications interface includes providing access to the medical data via a representational state transfer (REST) or a sockets application programming interface (API).

20 . The method of claim 14 wherein providing the at least one application comprises providing a medical device dashboard for graphically displaying the output indicative of the operating status of the medical devices.

21 . The method of claim 20 wherein the output indicative of the operating status includes an online status and a time when a last message was received from each of the medical devices.

22 . The method of claim 21 wherein the medical devices include medical devices of different types and from different medical device manufacturers.

23 . The method of claim 20 wherein the output indicative of the operating status includes an indication of a time between receipt of messages from a medical device.

24 . The method of claim 20 wherein the output indicative of the operating status includes an indication of a number of active medical devices connected to a server.

25 . The method of claim 14 wherein the anomaly detector is configured for monitoring usage patterns of the medical devices from the medical data, for comparing the usage patterns to usage patterns reconstructed based on models, and for indicating an anomaly if a monitored usage pattern differs from any of the reconstructed usage patterns by more than a threshold amount.

26 . The method of claim 25 comprising generating an anomaly score for a detected instance of an anomaly, determine whether the anomaly score satisfies a probably approximately correct (PAC) guarantee based on a false alarm rate (FAR) and a miss alarm rate (MAR), and output an anomaly prediction based on the PAC guarantee.

27 . A non-transitory computer readable medium having stored thereon executable instructions that when executed by at least one processor of a server control the server to perform steps comprising:

receiving, by a stream processor implemented on the at least one processor of the server, medical data from a plurality of medical devices and generating metadata and storing, in data storage implemented in hardware on the server, the medical data and the metadata for each of the devices, wherein the at least one processor of the server is implemented in hardware;

providing a communications interface implemented by the at least one processor on the server for implementing at least one communications protocol for accessing the medical data and the metadata;

providing at least one application for accessing the medical data and the metadata via the communications interface and generating, based on the medical data and the metadata, output indicative of operating status of the medical devices, wherein the at least one application includes an anomaly detector configured to monitor connection status of the medical devices, predict maintenance events concerning the medical devices as anomalies, prompt a user to confirm that the maintenance events occur as expected, and

responsive to receiving user input indicating that the maintenance events occur as expected, label data corresponding to the maintenance events as expected anomalies, generate a fine-grained calibration data set including the data corresponding to the maintenance events labeled as expected anomalies and data corresponding to other connection events as unexpected anomalies, and use the fine-grained calibration data set to adaptively recalibrate the anomaly detectors

wherein the at least one application includes a ventilation alert application, and

the steps further include extracting, by the ventilation alert application, in real time, and from messages sent by ventilators, an expired tidal volume variable value;

determining, by the ventilation alert application, whether the expired tidal volume variable value is greater than zero;

in response to determining that the expired tidal volume variable value is greater than zero, applying, by the ventilation alert application false positive detection logic that utilizes a patient name and a previous ventilator stop time to detect a false positive; and

when the false positive detection logic does not detect a false positive, sending, by the ventilation alert application, an alert to a virtual intensive care unit.