Personalized device recommendations for proactive health monitoring and management

1. A computer-implemented method for providing personalized recommendations of devices for monitoring and/or managing a health condition, the method comprising:

receiving, at a computer, first structured information regarding a patient and a first set of one or more patient populations;

receiving, at the computer, unstructured information regarding at least the patient and a second set of one or more patient populations, wherein the unstructured information comprises an electronic health record or portion(s) thereof;

analyzing, using the computer, the unstructured information to derive second structured information;

training a classification model using a training set comprising historical medical information for either or both of: the first set of one or more patient populations; and the second set of one or more patient populations, wherein the historical medical information comprises structured information and unstructured information, and wherein the classification model comprises a decision tree algorithm;

determining, using the computer, one or more health metrics to be monitored for the patient based on analyzing each of the first structured information and the second structured information using the classification model; and

determining, using the computer, an optimum set of devices to be used for monitoring the one or more health metrics;

wherein determining the optimum set of devices utilizes an optimization model comprising a plurality of rules configured to minimize a financial cost of the optimum set of devices while ensuring the optimum set of devices:

includes all device capabilities necessary to monitor the one or more health metrics; and

will not exceed applicable technological constraints of:

any respective one of the optimum set of devices; and

an operating environment in which the one or more health metrics are to be measured; and

wherein the rules comprise:

an objective function Σ i∈I c i ·X i configured to minimize a total cost of the optimum set of devices, wherein the total cost is defined by a plurality of constraints, comprising:

a capability constraint Σ i∈I z im ·X i ≥1, ∀m∈M configured to ensure the optimum set of devices includes all device capabilities necessary to monitor the one or more health metrics;

a resource constraint Σ i∈I a ij ·X i ≤K j , ∀j∈J configured to ensure the optimum set of devices will not exceed any of the applicable technological constraints of:

individual ones of the optimum set of devices; and

the operating environment in which the one or more health metrics are to be measured; and

a financial constraint Σ i∈I c i ·X i ≤B configured to ensure the financial cost of the optimum set of devices does not exceed a predetermined budget; and

wherein:

I is a set of possible devices to be included in the optimum set of devices;

c i is the financial cost of a given device i∈I;

B is a maximum budget for the optimum set of devices;

M is a set of measurements necessary to monitor the one or more health metrics;

z im is a binary input parameter having a value of 1 if the given device i∈I is capable of collecting measurement m∈M, and zero otherwise;

X i ∀i∈I is a binary variable having a value of 1 if the given device i∈I is one of the optimum set of devices recommended for monitoring and/or managing the health condition, and zero otherwise;

J is a set of resources available to the optimum set of devices in the operating environment in which the one or more health metrics are to be monitored;

a ij is a resource consumption of a given device i∈I; and

K j is an available capacity of resource j∈J.

2. The method as recited in claim 1 , wherein analyzing the unstructured information comprises building at least one extraction model configured to extract structured information from the unstructured information using a machine learning technique.

3. The method as recited in claim 1 , wherein the classification model also determines one or more weights associated with each of the one or more health metrics, the weights being indicative of relative importance of a given metric in predicting a future health status of the patient.

4. The method as recited in claim 1 , comprising building the decision tree algorithm, wherein building the decision tree algorithm comprises partitioning a decision tree model using one or more attributes based at least in part on: an entropy associated with the one or more attributes.

5. The method as recited in claim 1 , wherein the optimum set of devices is determined based on: device capabilities; associated device services; and resource constraints; and

wherein the resource constraints comprise financial constraints of the patient and technological constraints selected from the group consisting of: the device and an operating environment in which the health metrics are to be monitored for the patient.

6. The method as recited in claim 1 , comprising:

comparing values of the one or more health metrics of the patient to one or more corresponding event trigger thresholds;

upon determining, based on the comparison, that at least one of the one or more health metrics is characterized by a value exceeding the corresponding event trigger threshold, outputting the values of the one or more health metrics of the patient;

combining the values of the one or more health metrics of the patient with corresponding historical values of the one or more health metrics of the patient collected over time to generate a comprehensive set of values of the one or more health metrics of the patient;

determining a new set of one or more health metrics to be monitored for the patient based on analyzing the comprehensive set of values using the classification model; and

determining a new optimum set of devices to be used for monitoring the new set of one or more health metrics based on: device capabilities; associated device services; and resource constraints.

7. The method as recited in claim 1 , comprising assigning a plurality of labels to demographic data using a trained machine learning algorithm.

8. The method as recited in claim 7 , wherein the trained machine learning algorithm comprises a decision tree algorithm;

wherein the decision tree algorithm is characterized by including a root node and a plurality of leaf nodes; and

wherein each leaf node independently represents a set K of one or more most likely health condition(s) for patient(s) exhibiting a combination of demographics represented by a path from the root node to one of the leaf node(s).

9. The method as recited in claim 1 , wherein the decision tree algorithm is characterized by including a root node and a plurality of leaf nodes;

wherein each leaf node independently represents a set K of one or more most likely health condition(s) for patient(s) exhibiting a combination of demographics represented by a path from the root node to the leaf node; and

wherein the root node represents a demographic attribute associated with a particular health condition.

10. A computer program product for providing personalized recommendations of devices for monitoring and/or managing a health condition, the computer program product comprising a computer readable storage medium having program instructions embodied therewith, wherein the computer readable storage medium is not a transitory signal per se, the program instructions executable by a processor to cause the processor to perform a method comprising:

receiving, at the processor, first structured information regarding a patient and a first set of one or more patient populations;

receiving, at the processor, unstructured information regarding at least the patient and a second set of one or more patient populations wherein the unstructured information comprises an electronic health record or portion(s) thereof;

analyzing the unstructured information to derive second structured information;

training a classification model using a training set comprising historical medical information for either or both of: the first set of one or more patient populations; and the second set of one or more patient populations, wherein the historical medical information comprises structured information and unstructured information, and wherein the classification model comprises a decision tree algorithm;

determining, by the processor, one or more health metrics to be monitored for the patient based on analyzing the first structured information and the second structured information using the classification model; and

determining, by the processor, an optimum set of devices to be used for monitoring the one or more health metrics;

wherein determining the optimum set of devices utilizes an optimization model comprising a plurality of rules configured to minimize a financial cost of the optimum set of devices while ensuring the optimum set of devices:

includes all device capabilities necessary to monitor the one or more health metrics; and

will not exceed any applicable technological constraints of:

individual ones of the optimum set of devices; and

an operating environment in which the one or more health metrics are to be measured; and

wherein the rules comprise:

an objective function Σ i∈I c i ·X i configured to minimize a total cost of the optimum set of devices, wherein the total cost is defined by a plurality of constraints, comprising:

a capability constraint Σ i∈I z im ·X i ≥1, ∀m∈M configured to ensure the optimum set of devices includes all device capabilities necessary to monitor the one or more health metrics;

a resource constraint Σ i∈I a ij ·X i ≤K j , ∀∈J configured to ensure the optimum set of devices will not exceed any of the applicable technological constraints of:

the individual ones of the optimum set of devices; and

the operating environment in which the one or more health metrics are to be measured; and

a financial constraint Σ i∈I c i ·X i ≤B configured to ensure the financial cost of the optimum set of devices does not exceed a predetermined budget; and

wherein:

I is a set of possible devices to be included in the optimum set of devices;

c i is the financial cost of a given device i∈I,

B is a maximum budget for the optimum set of devices;

M is a set of measurements necessary to monitor the one or more health metrics,

z im is a binary input parameter having a value of 1 if wearable i∈I is capable of collecting measurement m∈M, and zero otherwise;

X i ∀i∈I is a binary variable having a value of 1 if the given device i∈I is one of the optimum set of devices recommended for monitoring and/or managing the health condition, and zero otherwise;

J is a set of resources available to the optimum set of devices in the operating environment in which the one or more health metrics are to be monitored;

a ij is a resource consumption of a given device i∈I; and

K j is an available capacity of resource j∈J.

11. The computer program product as recited in claim 10 , comprising building the decision tree algorithm, wherein building the decision tree algorithm comprises partitioning a decision tree model using one or more attributes based at least in part on: an entropy associated with the one or more attributes.

12. The computer program product as recited in claim 10 , wherein the classification model also determines one or more weights associated with each of the one or more health metrics, the weights being indicative of relative importance of a given metric in predicting a future health status of the patient.

13. The computer program product as recited in claim 10 , comprising generating a vector representing one or more probable causes of one or more health conditions based at least in part on the decision tree algorithm.

14. The computer program product as recited in claim 10 , wherein the optimum set of devices is determined based on: device capabilities; associated device services; and resource constraints; and

wherein the resource constraints comprise financial constraints of the patient and technological constraints selected from the group consisting of: the device and an operating environment in which the health metrics are to be monitored for the patient.

15. The computer program product as recited in claim 10 , comprising:

comparing values of the one or more health metrics of the patient to one or more corresponding event trigger thresholds;

upon determining, based on the comparison, that at least one of the one or more health metrics is characterized by a value exceeding the corresponding event trigger threshold, outputting the values of the one or more health metrics of the patient;

combining the values of the one or more health metrics of the patient with corresponding historical values of the one or more health metrics of the patient collected over time to generate a comprehensive set of values of the one or more health metrics of the patient;

determining a new set of one or more health metrics to be monitored for the patient based on analyzing the comprehensive set of values using the classification model; and

determining a new optimum set of devices to be used for monitoring the new set of one or more health metrics.

16. A computer-implemented method for providing personalized recommendations of devices for monitoring and/or managing a health condition, the method comprising:

receiving, at a computer, first structured information regarding a patient and a first set of one or more patient populations;

receiving, at the computer, unstructured information regarding at least the patient and a second set of one or more patient populations, wherein the unstructured information comprises an electronic health record or portion(s) thereof;

analyzing, using the computer, the unstructured information to derive second structured information;

training a classification model using a training set comprising historical medical information for either or both of: the first set of one or more patient populations; and the second set of one or more patient populations, wherein the historical medical information comprises structured information and unstructured information, and wherein the classification model comprises a decision tree algorithm;

determining, using the computer, one or more health metrics to be monitored for the patient based on analyzing each of the first structured information and the second structured information using the classification model; and

determining, using the computer, an optimum set of devices to be used for monitoring the one or more health metrics;

assigning a plurality of labels to demographic data using a trained machine learning algorithm;

generating a vector representing one or more probable causes of one or more health conditions based at least in part on the decision tree algorithm;

building the decision tree algorithm, wherein building the decision tree algorithm comprises partitioning a decision tree model using one or more attributes based at least in part on: an entropy associated with the one or more attributes; and/or an information gain associated with the one or more attributes;

wherein determining the optimum set of devices utilizes an optimization model comprising a plurality of rules configured to minimize financial cost of the optimum set of devices while ensuring the optimum set of devices:

includes all device capabilities necessary to monitor the one or more health metrics; and

will not exceed applicable technological constraints of:

any respective one of the optimum set of devices; and

an operating environment in which the one or more health metrics are to be measured; and

wherein the rules comprise an objective function Σ i∈I c i ·X i configured to minimize a total cost of the optimum set of devices, wherein the total cost is defined by a plurality of constraints, comprising:

a capability constraint Σ i∈I z im ·X i ≥1, ∀m∈M configured to ensure the optimum set of devices includes all device capabilities necessary to monitor the one or more health metrics;

a resource constraint Σ i∈I a ij ·X i ≤K j , ∀∈J configured to ensure the optimum set of devices will not exceed any of the applicable technological constraints of:

individual ones of the optimum set of devices; and

the operating environment in which the one or more health metrics are to be measured; and

a financial constraint Σ i∈I c i ·X i ≤B configured to ensure the financial cost of the optimum set of devices does not exceed a predetermined budget; and

wherein:

I is a set of possible devices to be included in the optimum set of devices;

c i is the financial cost of a given device i∈I;

B is a maximum budget for the optimum set of devices;

M is a set of measurements necessary to monitor the one or more health metrics;

z im is a binary input parameter having a value of 1 if the given device i∈I is capable of collecting measurement m∈M, and zero otherwise;

X i ∀i∈I is a binary variable having a value of 1 if the given device i∈I is one of the optimum set of devices recommended for monitoring and/or managing the health condition, and zero otherwise;

J is a set of resources available to the optimum set of devices in the operating environment in which the one or more health metrics are to be monitored;

a ij is a resource consumption of a given device i∈I; and

K j is an available capacity of resource j∈J; and

wherein the trained machine learning algorithm comprises a decision tree algorithm;

wherein the decision tree algorithm is characterized by including a root node and a plurality of leaf nodes;

wherein each leaf node independently represents a set K of one or more most likely health condition(s) for patient(s) exhibiting a combination of demographics represented by a path from the root node to the leaf node;

wherein the root node represents a demographic attribute associated with a particular health condition;

wherein the unstructured information comprises information selected from the group consisting of: images, audio, and videos;

wherein analyzing the unstructured information comprises building at least one extraction model configured to extract structured information from the unstructured information;

wherein the classification model also determines one or more weights associated with each of the one or more health metrics, the weights being indicative of relative importance of a given metric in predicting a future health status of the patient;

wherein the optimum set of devices is determined based on: device capabilities, associated device services, and resource constraints; and

wherein the resource constraints comprise financial constraints of the patient and technological constraints selected from the group consisting of: the device and an operating environment in which the health metrics are to be monitored for the patient.