Quality alerting for computer-aided dispatch systems
Examples provide a public safety answering point server including a processor configured to determines that a call taker of the PSAP has received a call from a caller regarding an incident and generate a transcript of the call. Based on the transcript, the processor determines a set of call attributes associated with the call, and determines whether a computer automated dispatch (“CAD”) record associated with the call is created within a threshold time period from the start time of the call. When the CAD record associated with the call is not created within the threshold time period, the processor infers, using a first artificial intelligence model, whether a CAD record should be created. In response to inferring that a CAD record should be created based on the set of call attributes, the processor generates and transmits a quality alert to an electronic device associated with a user of the PSAP.
1 . A public safety answering point (“PSAP”) server comprising:
an electronic processor; and
a non-transitory computer-readable medium comprising program instructions that, when executed by the electronic processor, cause the PSAP server to perform a set of operations comprising:
determining that a call taker of the PSAP has received a call from a caller regarding an incident;
generating a transcript of the call;
based on the transcript, determining a set of call attributes associated with the call;
determining whether a computer-aided dispatch (“CAD”) record associated with the call is created within a threshold time period from a start time of the call;
in response to determining that the CAD record associated with the call is not created within the threshold time period, inferring, using a first artificial intelligence (“AI”) model trained on historical data of transcripts of past calls and past CAD records, whether a CAD record associated with the call should be created based on the set of call attributes; and
in response to inferring that a CAD record associated with the call should be created based on the set of call attributes, generating and transmitting a quality alert to a supervisory device associated with a supervisor of the PSAP.
2 . The server of claim 1 , wherein the quality alert includes at least one selected from the group consisting of an identification of the call taker, a time of the call, the transcript of the call, a recording of the call, an identification of a CAD record associated with the call, and a time period between a start of the call and creation of a CAD record associated with the call.
3 . The server of claim 1 , wherein the set of call attributes include at least one selected from the group consisting of an incident type, a name of a person involved in the incident, a name of the caller, an organization, a location of the incident, a location of the caller, a date, a tone of the caller, an indication by the call taker that an emergency service unit will be dispatched, and instructions provided by the call taker to the caller.
4 . The server of claim 1 , wherein the set of operations further comprise:
in response to determining that the CAD record associated with the call is created within the threshold time period, determining a set of CAD record attributes of the CAD record associated with the call;
based on a comparison between a first subset of the call attributes and a first subset of the CAD attributes, inferring, using a second AI model trained on historical data of transcripts of past calls and past CAD records, whether emergency service units are sufficiently dispatched for the incident; and
in response to inferring that emergency service units are not sufficiently dispatched for the incident, generating and transmitting a quality alert to an electronic device associated with a user of the PSAP.
5 . The server of claim 4 , wherein inferring whether emergency service units are sufficiently dispatched includes at least one selected from the group consisting of inferring whether a sufficient type of emergency service unit is dispatched, inferring whether a sufficient quantity of emergency service units is dispatched for the incident, and inferring whether emergency service units are dispatched to a correct location.
6 . The server of claim 4 , wherein the set of operations further comprise determining a set of metadata of the call, the set of metadata including at least one selected from the group consisting of a location of the caller, a caller ID of the caller, a phone number of the caller, a time of the call, a duration of the call, a weather condition at the location of the caller, and a set of historical metadata and call attributes associated with past calls from the caller; and
inferring, using the second AI model, whether emergency service units are sufficiently dispatched for the incident based on a comparison between a first subset of metadata and a second subset of CAD record attributes.
7 . The server of claim 4 , wherein the set of operations further comprise providing the transcript of the call and the CAD record as training data to the second AI model.
8 . The server of claim 1 , wherein the threshold time period is a predetermined time period.
9 . The server of claim 1 , wherein the historical data includes a historical distribution of past CAD record creation time periods, each past CAD record creation time period being a time period between a past receiving of a call by an emergency service provider and creation of a corresponding CAD record, and
the set of operations further comprise determining the threshold time period based on the historical distribution.
10 . The server of claim 9 , wherein the electronic processor is configured to determine the threshold time period according to a predetermined percentile of the historical distribution.
11 . The server of claim 9 , wherein the set of call attributes include an incident type, and the historical distribution is conditioned on the incident type.
12 . The server of claim 1 , wherein the data is based on data from a plurality of PSAPs.
13 . The server of claim 1 , wherein the electronic processor determines the threshold time period using the first AI model.
14 . The server of claim 1 , wherein the set of call attributes include an incident type, and the incident type includes at least one selected from the group consisting of a medical emergency, a mass casualty event, a theft, a domestic dispute, a break-in, a traffic accident, a robbery, an assault, a public threat, a mental health crisis, a weather-related incident, a fire, an administrative question from the caller, a follow-up regarding an active incident, and an accidental call.
15 . A public safety answering point (“PSAP”) server comprising:
an electronic processor; and
a non-transitory computer-readable medium comprising program instructions that, when executed by the electronic processor, cause the PSAP server to perform a set of operations comprising:
determining that a call taker of the PSAP has received a call from a caller regarding an incident;
generating a transcript of the call;
based on the transcript, determining a set of call attributes associated with the call;
identifying a computer-aided dispatch (“CAD”) record associated with the call is created;
determining a set of CAD record attributes of the CAD record associated with the call;
based on a comparison between a first subset of the call attributes and a first subset of the CAD attributes, inferring, using a first AI model trained on historical data of transcripts of past calls and past CAD records, whether emergency service units are sufficiently dispatched for the incident; and
in response to inferring that emergency service units are not sufficiently dispatched for the incident, generating and transmitting a quality alert to a supervisory device associated with a supervisor of the PSAP.
16 . The server of claim 15 , wherein the quality alert includes at least one selected from the group consisting of an identification of the call taker, a time of the call, the transcript of the call, a recording of the call, an identification of a CAD record associated with the call, and a time period between a start of the call and creation of a CAD record associated with the call.
17 . The server of claim 15 , wherein the set of call attributes include at least one selected from the group consisting of an incident type, a name of a person involved in the incident, a name of the caller, an organization, a location of the incident, a location of the caller, a date, a time, a tone of the caller, an indication by the call taker that an emergency service unit will be dispatched, an instructions provided by the call taker to the caller.
18 . The server of claim 15 , wherein inferring whether emergency service units are sufficiently dispatched includes at least one selected from the group consisting of inferring whether a sufficient type of emergency service unit is dispatched, inferring whether a sufficient quantity of emergency service units is dispatched for the incident, and inferring whether emergency service units are dispatched to a correct location.
19 . The server of claim 15 , wherein the set of operations further comprise:
determining whether the CAD record associated with the call is created within a threshold time period from a start time of the call;
in response to determining that the CAD record associated with the call is not created within the threshold time period, inferring, using a second AI model trained on historical data of transcripts of past calls and past CAD records, whether a CAD record associated with the call should be created based on the set of call attributes; and
in response to inferring that a CAD record associated with the call should be created based on the set of call attributes, generating and transmitting a quality alert to an electronic device associated with a user of the PSAP.
20 . A method for a public safety answering point (“PSAP”) server, the method comprising:
determining that a call taker of the PSAP has received a call from a caller regarding an incident;
generating a transcript of the call;
based on the transcript, determining a set of call attributes associated with the call;
determining whether a computer-aided dispatch (“CAD”) record associated with the call is created within a threshold time period from a start time of the call;
in response to determining that the CAD record associated with the call is not created within the threshold time period, inferring, using a first artificial intelligence (“AI”) model trained on historical data of transcripts of past calls and past CAD records, whether a CAD record associated with the call should be created based on the set of call attributes; and
in response to inferring that a CAD record associated with the call should be created based on the set of call attributes, generating and transmitting a quality alert to a supervisory device associated with a supervisor of the PSAP.