Systems and methods for disease progression modeling

1. A method for determining a disease state transition path of glaucoma, said method comprising:

receiving, by a first processor of a disease progression model, patient data comprising structural degeneration data and functional degeneration data related to a patient, via a secure interface, from a second processor associated with a healthcare provider terminal;

identifying, by the first processor, based on the patient data, a first disease state of a plurality of non-overlapping disease states of glaucoma each associated with a predetermined range of one or both of functional and structural degeneration values;

identifying, by the first processor, based on the patient data, a second disease state of the plurality of disease states, wherein the second disease state is non-adjacent to the first disease state;

predicting, by the first processor using a two dimensional continuous-time hidden Markov model, a most probable path between the first disease state and the second disease state, wherein the most probable path comprises one or more intermediary disease states of the plurality of disease states, wherein each intermediary disease state is adjacent to one or more of the first disease state, the second disease state, and another intermediary disease state;

outputting, by the first processor, the most probable path between the first disease state and the second disease state to cause the healthcare provider terminal to display the most probable path as a graphical user interface;

training, by the first processor, the hidden Markov model based on patient progression data corresponding to a plurality of patients having an attribute;

predicting, with the first processor using a trained hidden Markov model, an updated most probable path between the first disease state and the second disease state;

predicting, with the first processor, a disease progression rate based on the updated most probable path;

comparing, by the first processor, the predicted disease progression rate with a predetermined disease progression rate; and

predicting, with the first processor, whether a patient has the attribute based on the comparison.

2. The method of claim 1 , wherein each sequential disease state is associated with one or both of increased structural and functional degeneration values as time progresses.

3. The method of claim 1 further comprising iteratively updating, by the first processor, one or more parameters of the hidden Markov model based on the determined most probable path until the most probable path remains substantially constant.

4. The method of claim 1 , wherein predicting the most probable path further comprises:

predicting the most probable path, with the first processor using hidden Markov model and assigning, by the first processor, a uniform time spent in each intermediary state along the most probable path;

updating, by the first processor, one or more parameters of the hidden Markov model based on the determined most probable path;

predicting, by the first processor, the most probable path using the hidden Markov model; and

alternating, by the first processor, the updating and redetermining steps until the redetermined most probable path substantially matches a previously determined most probable path.

5. The method of claim 1 further comprising predicting, with the first processor, a most probable next disease state for one or more disease states along the most probable path or for the second disease state using the hidden Markov model.

6. The method of claim 5 further comprising:

comparing, by the first processor, the most probable next disease state with the second disease state; and

predicting, with the first processor, that a transition between disease states along the most probable path has a fast structural and/or functional progression based on the comparison.

7. The method of claim 5 further comprising predicting, with the first processor, an expected time to transition from the second disease state to the most probable next disease state using the hidden Markov model.

8. A method for detecting disease state transitions of glaucoma, said method comprising:

receiving, by a first processor of a disease progression model, patient data comprising structural degeneration data and functional degeneration data related to a patient, via a secure interface, from a second processor associated with a healthcare provider terminal;

identifying, by the first processor, based on the patient data, a first disease state of a plurality of non-overlapping disease states of glaucoma each associated with a predetermined range of one or both of functional and structural degeneration values;

identifying, by the first processor, based on the patient data, a second disease state of the plurality of disease states, wherein the second disease state is non-adjacent to the first disease state;

determining, with the first processor using a two dimensional continuous-time hidden Markov model, a most probable path between the first disease state and the second disease state, wherein:

the most probable path comprises one or more intermediary disease states of the plurality of disease states, and

each intermediary disease state is adjacent to one or more of the first disease state, the second disease state, and another intermediary disease state;

determining, with the first processor, a most probable next disease state for one or more disease states along the most probable path using the hidden Markov model;

comparing, by the first processor, the most probable next disease state with the second disease state;

determining, with the first processor, that a transition between disease states along the most probable path has a fast structural and/or functional progression based on the comparison;

when the transition is determined to be the fast structural and/or functional progression, training, by the first processor, the hidden Markov model based on patient progression data corresponding to a plurality of patients having the fast structural and/or functional progression; and

redetermining, with the first processor, the most probable next disease state for one or more disease states along the most probable path using the hidden Markov model.

9. The method of claim 8 , wherein each sequential disease state is associated with one or both of increased functional and structural degeneration values as time progresses.

10. The method of claim 8 further comprising iteratively updating, by the first processor, one or more parameters of the hidden Markov model based on the determined most probable path until the most probable path remains substantially constant.

11. A system for determining a disease state transition path of glaucoma, said method comprising:

a storage device for storing instructions; and

a processor configured to execute the instructions in the storage device to:

receive patient data comprising structural degeneration data and functional degeneration data related to a patient;

identify, based on the patient data, two or more disease states of a plurality of nonoverlapping disease states of glaucoma each associated with a predetermined range of one or both of functional and structural degeneration values;

predict, using a two dimensional continuous-time hidden Markov model, a most probable path between a sequential pair of non-adjacent disease states of the two or more disease states, wherein the most probable path comprises one or more intermediary disease states of the plurality of disease states,

wherein each intermediary disease state is adjacent to at least one of the sequential pair of non-adjacent disease states and another intermediary disease state;

train the hidden Markov model based on patient progression data corresponding to a plurality of patients having an attribute;

predict, using a trained hidden Markov model, an updated most probable path between the sequential pair of non-adjacent disease states;

predict a disease progression rate based on the updated most probable path;

compare the predicted disease progression rate with a predetermined disease progression rate; and

determine whether a patient has the attribute based on the comparison.

12. The system of claim 11 , wherein each of the identified two or more disease states is associated with one or both of increased functional and structural degeneration values as time progresses.

13. The system of claim 11 , wherein the processor is further configured to predict a most probable future disease state for the most recent of the identified two or more disease states using the hidden Markov model.

14. The system of claim 13 , wherein the processor is further configured to predict an expected time to transition from the most recent of the identified two or more disease state to the most probable future disease state using the hidden Markov model.