NATURAL-LANGUAGE TEXT GENERATION WITH PANDEMIC-BIO-SURVEILLANCE MULTI PATHOGEN SYSTEMS

1 . A tangible, non-transitory, machine-readable medium storing instructions that, when executed by one or more processors, effectuate operations comprising:

obtaining, with a computer system, identifiers of a set of leaders of an organization and identifiers of a set of participants of the organization;

obtaining, with the computer system, a set of geolocations that are respective facilities of the organization;

obtaining, with the computer system, geolocation-pathogen-risk scores of the geolocations that are respective facilities of the organization;

obtaining, with the computer system, a current, first state of a finite state machine, the finite state machine having four or more states, each of the four or more states corresponding to a level of at least some of the geolocation-pathogen-risk scores or a first derivative with respect to time of at least some of the geolocation-pathogen-risk scores;

determining, with the computer system, to transition to a second state of the finite state machine based on the geolocation-pathogen-risk scores of the geolocations that are respective facilities of the organization;

executing, with the computer system, logic of the second state configured to cause a user interface to be presented on computing devices of at least some of the leaders of the organization, the user interface including:

a characterization of the at least some of the geolocation-pathogen-risk scores of the geolocations that are respective facilities of the organization, and

one or more proposed natural language text messages to at least some of the participants of the organization;

receiving, with the computer system, a selection or modification of at least one of the one or more proposed natural language text messages to at least some of the participants of the organization; and

causing, with the computer system, the selected or modified at least one of the natural language text messages to be presented to the at least some of the participants of the organization.

2 . The medium of claim 1 , the operations further comprising:

obtaining a plurality of geolocations, including the geolocations of the facilities, from a geographic information system, the plurality of geolocations including both geolocations that are geographic regions and geolocations that are places of interest within those geographic regions;

obtaining first data about at least some of the plurality of geolocations, wherein:

the first data is updated less frequently than a first rate, and

the first data is not pathogen specific;

obtaining second data about at least some of the plurality of geolocations, wherein:

the second data is updated more frequently than a second rate that is more frequent than the first rate, and

the second data is pathogen-specific;

determining geolocation-pathogen-risk scores of the geographic regions based on both the first data and the second data;

determining geolocation-pathogen-risk scores of the places of interest with a machine learning model trained to allocate risk of geographic regions to places of interest within those geographic regions based on at least some of the first data, such that at least some places of interest in the same geographic region have different geolocation-pathogen-risk scores; and

storing the geolocation-pathogen-risk scores of the places of interest and the geolocation-pathogen-risk scores of the geographic regions in memory, wherein:

the first data is static data;

the second data is dynamic data;

at least some of the geographic regions are reporting districts;

at least some of the places of interest are smaller than 10,000 square meters;

the geolocation-pathogen-risk scores of the places of interest and the geolocation-pathogen-risk scores of the geographic regions are specific to the corresponding geolocations and are independent of potential visitors to those corresponding geolocations;

the static data includes census data updated once per decade; and

the dynamic data includes amounts of infections or deaths attributable to a given pathogen in the geographic regions updated at least daily.

3 . The medium of claim 2 , the operations further comprising:

obtaining, for each of at least some members of the set of participants, associated geolocations, at least some of the associated geolocations being geolocations including residences of respective members of the set or participants or places the respective members of the set of participants have previously visited;

accessing, for each of the at least some members of the set of participants, the geolocation-pathogen-risk scores of the associated geolocations; and

determining geolocation-pathogen-risk scores, or modified geolocation-pathogen-risk scores, of the facilities based on the geolocation-pathogen-risk scores of the associated geolocations.

4 . The medium of claim 3 , the operations further comprising:

determining subsets of the set of people based on the geolocation-pathogen-risk scores of their associated geolocations.

5 . The medium of claim 3 , the operations further comprising:

causing a message to be sent to computing devices of one of the subsets having the most severe geolocation-pathogen-risk score among the subsets, the message instructing the members of the one of the subsets to not visit the facilities; and

causing the user interface to be presented showing color-coded representations of the facilities with colors indicating the geolocation-pathogen-risk scores of the facilities.

6 . The medium of claim 3 , wherein obtaining the set of participants and obtaining the associated geolocations comprises querying a payroll system or enterprise resource planning system of the organization to obtain residential addresses of employees of the organization.

7 . The medium of claim 1 , wherein the four or more states of the finite state machine include at least one state corresponding to the first derivative with respect to time of at least one of the geolocation-pathogen-risk scores of the geolocations that are respective facilities of the organization.

8 . The medium of claim 1 , wherein the four or more states of the finite state machine include at least two states corresponding to the first derivative with respect to time of at least one of the geolocation-pathogen-risk scores of the geolocations that are respective facilities of the organization, one of the at least two states corresponding to a higher first derivative and another of the at least two states corresponding to a lower first derivative that is lower than the higher first derivative.

9 . The medium of claim 8 , wherein the operations further comprise:

determining that the first derivative with respect to time of at least one of the geolocation-pathogen-risk scores of the geolocations that are respective facilities of the organization is greater than a threshold and, in response, transitioning between states of the finite state machine.

10 . The medium of claim 1 , wherein the at least four states comprise the following states:

a first state corresponding to a lower range of geolocation-pathogen-risk scores of at least some of the geolocations that are respective facilities of the organization;

a second state corresponding to a higher range of geolocation-pathogen-risk scores of at least some of the geolocations that are respective facilities of the organization, the higher range of geolocation-pathogen-risk scores being higher than the lower range of geolocation-pathogen-risk scores;

a third state corresponding to the lower first derivative with respect to time of geolocation-pathogen-risk scores of at least some of the geolocations that are respective facilities of the organization; and

a fourth state corresponding to a higher first derivative with respect to time of geolocation-pathogen-risk scores of at least some of the geolocations that are respective facilities of the organization, the higher first derivative being higher than the lower first derivative.

11 . The medium of claim 1 , wherein the finite state machine is instantiated for each of the facilities, each such instance being configured to have a different state from the others.

12 . The medium of claim 1 , wherein executing logic of the second state comprises:

determining a topic of message based on the transition to the second state; and

determining, by a natural-language-text generation model, natural language text within the topic of the message.

13 . The medium of claim 12 , wherein the natural language text is a warning of risk of infection from a pathogen at least one of the facilities, and wherein the operations further comprise:

obtaining feedback indicating whether participants complied with an instruction expressed by the natural language text; and

adjusting parameters of the natural-language-text generation model based on the feedback to reduce a likelihood of the natural language text being presented again.

14 . The medium of claim 12 , wherein the natural-language-text generation model comprises a reinforcement learning model trained to sequentially determine natural language text messages to present as candidate messages in the user interface based on feedback indicating compliance or lack of compliance with previous messages presented to participants, the reinforcement learning model having a value function that penalizes semantic similarity between messages within a threshold sequence position-difference in a sequence of messages.

15 . The medium of claim 1 , wherein the computer system implements a multi-tenant software-as-a-service application configured to cause instances of the user interface to be presented to leadership of a plurality of different organizations.

16 . The medium of claim 1 , the operations further comprising:

receiving a modification, input via the user interface, to one or more proposed natural language text messages and, in response, scoring the modification based on predicted likelihood of compliance by at least some of the participants; and

updating the user interface to present an indication of the score based on predicted likelihood of compliance by at least some of the participants.

17 . The medium of claim 1 , the operations further comprising:

steps for organizational messaging.

18 . The medium of claim 1 , the operations further comprising:

steps for reducing latency when querying a geographic information system;

steps for calculating geolocation-pathogen-risk scores;

steps for personal-pathogen-risk-scoring; and

steps for behavior modification messaging.

19 . The medium of claim 1 , wherein:

the operations further comprise:

obtaining a plurality of geolocations, including the geolocations of the facilities, from a geographic information system, the plurality of geolocations including both geolocations that are geographic regions and geolocations that are places of interest within those geographic regions;

obtaining first data about at least some of the plurality of geolocations, wherein:

the first data is updated less frequently than a first rate, and

the first data is not pathogen specific;

obtaining second data about at least some of the plurality of geolocations, wherein:

the second data is updated more frequently than a second rate that is more frequent than the first rate, and

the second data is pathogen-specific;

determining geolocation-pathogen-risk scores of the geographic regions based on both the first data and the second data;

determining geolocation-pathogen-risk scores of the places of interest with a machine learning model trained to allocate risk of geographic regions to places of interest within those geographic regions based on at least some of the first data, such that at least some places of interest in the same geographic region have different geolocation-pathogen-risk scores;

storing the geolocation-pathogen-risk scores of the places of interest and the geolocation-pathogen-risk scores of the geographic regions in memory, wherein:

the first data is static data;

the second data is dynamic data;

at least some of the geographic regions are reporting districts;

at least some of the places of interest are smaller than 10,000 square meters;

the geolocation-pathogen-risk scores of the places of interest and the geolocation-pathogen-risk scores of the geographic regions are specific to the corresponding geolocations and are independent of potential visitors to those corresponding geolocations;

the static data includes census data updated once per decade; and

the dynamic data includes amounts of infections or deaths attributable to a given pathogen in the geographic regions updated at least daily;

obtaining, for each of at least some members of the set of participants, associated geolocations, at least some of the associated geolocations being geolocations including residences of respective members of the set or participants or places the respective members of the set of participants have previously visited;

accessing, for each of the at least some members of the set of participants, the geolocation-pathogen-risk scores of the associated geolocations;

determining geolocation-pathogen-risk scores, or modified geolocation-pathogen-risk scores, of the facilities based on the geolocation-pathogen-risk scores of the associated geolocations;

determining subsets of the set of people based on the geolocation-pathogen-risk scores of their associated geolocations;

causing a message to be sent to computing devices of one of the subsets having the most severe geolocation-pathogen-risk score among the subsets, the message instructing the members of the one of the subsets to not visit the facilities; and

causing the user interface to be presented showing color-coded representations of the facilities with colors indicating the geolocation-pathogen-risk scores of the facilities;

obtaining the set of participants and obtaining the associated geolocations comprises querying a payroll system or enterprise resource planning system of the organization to obtain residential addresses of employees of the organization;

the four or more states of the finite state machine include at least one state corresponding to the first derivative with respect to time of at least one of the geolocation-pathogen-risk scores of the geolocations that are respective facilities of the organization;

the four or more states of the finite state machine include at least two states corresponding to the first derivative with respect to time of at least one of the geolocation-pathogen-risk scores of the geolocations that are respective facilities of the organization, one of the at least two states corresponding to a higher first derivative and another of the at least two states corresponding to a lower first derivative that is lower than the higher first derivative;

the operations further comprise determining that the first derivative with respect to time of at least one of the geolocation-pathogen-risk scores of the geolocations that are respective facilities of the organization is greater than a threshold and, in response, transitioning between states of the finite state machine;

the at least four states comprise the following states:

a first state corresponding to a lower range of geolocation-pathogen-risk scores of at least some of the geolocations that are respective facilities of the organization;

a second state corresponding to a higher range of geolocation-pathogen-risk scores of at least some of the geolocations that are respective facilities of the organization, the higher range of geolocation-pathogen-risk scores being higher than the lower range of geolocation-pathogen-risk scores;

a third state corresponding to the lower first derivative with respect to time of geolocation-pathogen-risk scores of at least some of the geolocations that are respective facilities of the organization; and

a fourth state corresponding to a higher first derivative with respect to time of geolocation-pathogen-risk scores of at least some of the geolocations that are respective facilities of the organization, the higher first derivative being higher than the lower first derivative;

the finite state machine is instantiated for each of the facilities, each such instance being configured to have a different state from the others;

executing logic of the second state comprises:

determining a topic of message based on the transition to the second state; and

determining, by a natural-language-text generation model, natural language text within the topic of the message;

the natural language text is a warning of risk of infection from a pathogen at least one of the facilities, and wherein the operations further comprise:

obtaining feedback indicating whether participants complied with an instruction expressed by the natural language text; and

adjusting parameters of the natural-language-text generation model based on the feedback to reduce a likelihood of the natural language text being presented again;

the natural-language-text generation model comprises a reinforcement learning model trained to sequentially determine natural language text messages to present as candidate messages in the user interface based on feedback indicating compliance or lack of compliance with previous messages presented to participants, the reinforcement learning model having a value function that penalizes semantic similarity between messages within a threshold sequence position-difference in a sequence of messages;

the computer system implements a multi-tenant software-as-a-service application configured to cause instances of the user interface to be presented to leadership of a plurality of different organizations; and

the operations further comprise:

receiving a modification, input via the user interface, to one or more proposed natural language text messages and, in response, scoring the modification based on predicted likelihood of compliance by at least some of the participants;

updating the user interface to present an indication of the score based on predicted likelihood of compliance by at least some of the participants;

steps for organizational messaging;

steps for reducing latency when querying a geographic information system;

steps for calculating geolocation-pathogen-risk scores;

steps for personal-pathogen-risk-scoring; and

steps for behavior modification messaging.

20 . A method, comprising:

obtaining, with a computer system, identifiers of a set of leaders of an organization and identifiers of a set of participants of the organization;

obtaining, with the computer system, a set of geolocations that are respective facilities of the organization;

obtaining, with the computer system, geolocation-pathogen-risk scores of the geolocations that are respective facilities of the organization;

obtaining, with the computer system, a current, first state of a finite state machine, the finite state machine having four or more states, each of the four or more states corresponding to a level of at least some of the geolocation-pathogen-risk scores or a first derivative with respect to time of at least some of the geolocation-pathogen-risk scores;

determining, with the computer system, to transition to a second state of the finite state machine based on the geolocation-pathogen-risk scores of the geolocations that are respective facilities of the organization;

executing, with the computer system, logic of the second state configured to cause a user interface to be presented on computing devices of at least some of the leaders of the organization, the user interface including:

a characterization of the at least some of the geolocation-pathogen-risk scores of the geolocations that are respective facilities of the organization, and

one or more proposed natural language text messages to at least some of the participants of the organization;

receiving, with the computer system, a selection or modification of at least one of the one or more proposed natural language text messages to at least some of the participants of the organization; and

causing, with the computer system, the selected or modified at least one of the natural language text messages to be presented to the at least some of the participants of the organization.