Application US20200152330

Scalable Personalized Treatment Recommendation
[+] Show Details

US20200152330

1. A method comprising: (6) (5)

generating, using one or more processors, a personalized respiratory health profile for each user of a plurality of users, each profile including information collected from monitoring the user's respiratory symptoms over time;

clustering, using the one or more processors, the generated profiles into one or more categories using clustering techniques;

associating, using the one or more processors, each category with a set of actions;

clustering, using the one or more processors, a particular user into one or more of the categories based on the particular user's profile; and

recommending, using the one or more processors, one or more actions for the particular user from one or more sets of actions for the one or more categories.

2. The method of claim 1, wherein the information includes at least one of the user's respiratory symptoms, disease states, disease trends over time, location, age, compliance with a treatment plan or triggers. (0)

4. The method of claim 1, wherein at least one category includes users who share weak or strong compliance to medical treatment, users having a same postal code or users with similar symptoms, or other hidden categories discovered by a network. (0)

7. The method of claim 1, wherein the set of actions includes at least one of guidance on medications, treatment plans, or trigger avoidance. (0)

8. The method of claim 1, further comprising: (3) (3)

obtaining a set of feature vectors for each user over a period of time;

predicting, using a pre-trained Long Short-Term Memory (LSTM) network or Gated Recurrent Unit (GRU) network, the user's disease or disease state; and

recommending one or more actions to the user to prevent or lower a probability of the user being in the predicted diseased state in the future.

9. The method of claim 8, wherein the LSTM network or GRU network is trained as a sequence-to-sequence network using a user feature vector at each timestamp in the past and the disease states in the past as input to the LSTM network or GRU network, and the user's disease state from present time as an output. (1) (0)

25. The method of claim 8, wherein one or more features from a timestamp that are not available are replaced by the user's most recent value for that feature or left blank. (0)

26. The method of claim 8, wherein a probability of the user's future disease state is predicted based on available past, present, and future feature prediction. (1) (0)

21. The method of claim 1, wherein supervised or unsupervised clustering methods are used to find patterns that cluster the plurality of users into the one or more categories. (1) (0)

23. The method claim 1, wherein the plurality of users get assigned a new cluster or clusters and receive a more personalized set of actions over time. (1) (0)

3. (canceled) (0)

5. (canceled) (0)

6. (canceled) (0)

11. A system comprising: (8) (6)

one or more processors;

memory storing instructions that when executed by the one or more processors, cause the one or more processors to perform operations comprising:

generating a personalized respiratory health profile for each user of a plurality of users, each profile including information collected from monitoring the user's respiratory symptoms over time;

clustering the generated profiles into one or more categories using clustering techniques;

associating each category with a set of actions;

clustering a particular user into one or more categories based on the particular user's profile; and recommending one or more actions for the particular user from one or more sets of actions for the one or more categories.

12. The system of claim 11, wherein the information includes at least one of the user's respiratory symptoms, disease states, trends, location, age, compliance or triggers. (0)

14. The system of claim 11, wherein at least one category includes users who share weak or strong compliance to medical treatment, users having a same postal code or users with similar symptoms. (0)

15. The system of claim 11, wherein the profiles are updated over time. (0)

16. The system of claim 11, wherein the categories are updated periodically and each category contains a different set of user profiles. (0)

17. The system of claim 11, where in the set of actions includes at least one of guidance on medications, treatment plans, or trigger avoidance. (0)

18. The system of claim 11, further comprising: (3) (3)

obtaining a set of feature vectors for each user over a period of time;

predicting, using a pre-trained Long Short-Term Memory (LSTM) network or Gated Recurrent Unit (GRU) network, the user's disease or disease state;

recommending one or more actions to the user to prevent or lower a probability of the user being in the diseased state in the future.

19. The system of claim 18, wherein the LSTM network or GRU network is trained as a sequence-to-sequence network using a user feature vector at each timestamp in the past and the disease states in the past as input to the LSTM network or GRU network, and the user's disease state from present time as an output. (1) (0)

32. The system of claim 18, wherein one or more features from a timestamp that are not available are replaced by the user's most recent value for that feature or left blank. (0)

33. The system of claim 18, wherein a probability of the user's future disease state is predicted based on available past, present, and future feature prediction. (1) (0)

28. The system of claim 11, wherein supervised or unsupervised clustering methods are used to find patterns that cluster the plurality of users into the one or more categories. (1) (0)

30. The system of claim 11, wherein the user is assigned a new cluster or clusters and receives a more personalized set of actions over time. (1) (0)

13. (canceled) (0)

View Abstract and Specification Size

PDF with Images and Document Face >

Full Text Publication >

Application US20200152330

Scalable Personalized Treatment Recommendation

Patent Matrix® Search