Back to results
US 11,232,870 B1Granted
Communicable disease prediction and control based on behavioral indicators derived using machine learning
AssigneeNeura Labs Corp.
InventorsTriinu Magi Shaashua, Amit Hammer, Ori Shaashua
Life of this patent
Estimating the term…
Abstract
Methods, apparatus, and systems for predicting and controlling communicable diseases are disclosed. In one example aspect, a method for predicting a communicable disease includes receiving, for each member of a community, multiple data streams associated with the member from multiple sensor devices, and computing, based on a social or locational relationship between the member and other entities in the community and a timeline of activities performed by the member according to the timestamp for each data packet, a list of behavioral indicators for the community indicating a current state of the communicable disease using one or more machine learning models.
Claims28
A computer-implemented method for tracking and predicting a communicable disease, comprising:
- sampling data packets from multiple data streams associated with a member from multiple devices at a predefined frequency, wherein the multiple devices include at least one connectivity device configured to provide network connection to at least one of remaining multiple devices, and wherein a data packet in each of the multiple data streams is associated with a timestamp indicating a time at which the data packet is sampled from a corresponding device;
- validating the data packets by discarding data packets that have invalid data fields, wherein an invalid data field is determined according to a type of a data stream associated with the data packet;
- determining a locational entity associated with the member based on how the data packets of the multiple data streams are routed by the at least one connectivity device of the multiple devices,
- determining, using at least one machine learning model that comprises an ensemble of decision trees, a community associated with the locational entity of the member based on the multiple data streams, wherein there exists a social or locational relationship between the member and other entities in the community, and wherein the at least one machine learning model is trained by reinforcing disease-transmission features that are correlated among the ensemble of decision trees;
- and
- computing, using one or more additional machine learning models based on the social or locational relationship between the member and other entities in the community and a timeline of activities performed by the member according to the timestamp for each data packet, a list of behavioral indicators for the community, wherein the list of behavioral indicators and a corresponding set of epidemiological data of the community indicate a current state of the communicable disease in the community.
3 more elements
A system for predicting a transmission trend of a communicable disease, comprising one or more processors configured to:
- receive a set of epidemiological data of the communicable disease;
- sample, for each member of a community, data packets from multiple data streams associated with the member from multiple devices at a predefined frequency, wherein the multiple devices include at least one connectivity device configured to provide network connection to at least one of remaining multiple devices, and wherein the sampled data packets of the multiple data streams are validated consolidated to form behavioral timelines of members in the community;
- determine at least one locational entity associated with each member based on how the data packets of the multiple data streams are routed by the at least one connectivity device,
- determine, using at least one machine learning model that comprises an ensemble of decision trees, the community associated with the at least one locational entity of each member based on the multiple data streams, wherein there exists a social or locational relationship between the member and other entities in the community, and wherein the at least one machine learning model is trained by reinforcing disease-transmission features that are correlated among the ensemble of decision trees;
- determine, using one or more additional machine learning models, a list of behavioral indicators that indicates a current state of the communicable disease in the community based on the behavioral timelines of members in the community;
- and
- predict the transmission trend of the communicable disease within the community based on the list of behavioral indicators and the set of epidemiological data.
4 more elements
A system for determining transmission risks of a communicable disease in a community, comprising one or more processors configured to:
- sample, for each member of the community, data packets from multiple data streams associated with the member from multiple devices at a predefined frequency, wherein the multiple devices include at least one connectivity device configured to provide network connection to at least one of remaining multiple devices, and wherein the multiple data streams are validated and consolidated to form behavioral timelines of members in the community;
- determine at least one locational entity associated with each member based on how the data packets of the multiple data streams are routed by the at least one connectivity device;
- determine, using at least one machine learning model that comprises an ensemble of decision trees, the community associated with the at least one locational entity of each member based on the multiple data streams, wherein there exists a social or locational relationship between the member and other entities in the community, and wherein the at least one machine learning model is trained by reinforcing disease-transmission features that are correlated among the ensemble of decision trees;
- determine, using one or more additional machine learning models, a list of behavioral indicators for the community based on the behavioral timelines of the members in the community;
- and
- generate at least one risk score for the community based on the list of behavioral indicators.
3 more elements
A system for performing test pooling of a communicable disease in a community, comprising one or more processors configured to:
- sample, for each member of the community, data packets from multiple data streams associated with the member from multiple devices at a predefined frequency, wherein the multiple devices include at least one connectivity device configured to provide network connection to at least one of remaining multiple devices, and wherein the multiple data streams are validated and consolidated to form behavioral timelines of members in the community;
- determine at least one locational entity associated with each member based on how the data packets of the multiple data streams are routed by the at least the connectivity device of the multiple devices,
- determine, using at least one machine learning model that comprises an ensemble of decision trees, the community associated with the at least one locational entity of each member based on the multiple data streams, wherein there exists a social or locational relationship between the member and other entities in the community, and wherein the at least one machine learning model is trained by reinforcing disease-transmission features that are correlated among the ensemble of decision trees;
- determine, using one or more additional machine learning models, a list of behavioral indicators for the community based on the behavioral timelines of the members in the community;
- collect test samples taken from a subset of the members of the community;
- determine, based on the list of behavioral indicators for the community, a risk score associated with each of the test samples;
- and
- determine, based on the risk score associated with each of the test samples, a number of samples to be pooled together in a single test run for the communication disease.
5 more elements
Dependencies are inferred from the claim text.