Integrated system and method for personalized stratification and prediction of neurodegenerative disease

An integrated system for predicting Neurodegenerative Disease (ND) progression rate, the system comprising:

• an input device operative to receive feature-based patient representations;
• a computing device configured to:
• receive feature-based representation of ND patients;
• hierarchically cluster individual Functionality Measure (FM) items into group FM items, the clustering implemented on a basis of most closely correlating individual FM item values;
• group individual FM items into body function groups in accordance with group FM items;
• calculate group FM item progression rates for each of the body function groups;
• establish a multi-dimensional patient representation based on the group FM item progression rates;
• cluster the multi-dimensional patient representation into distinct patient clusters;
• identify an optimal number of patient clusters from among patient data in accordance with a data driven optimization scheme;
• assign a new patient to one of the patient clusters in accordance with a best match between a progression rate of the new patient and a progression rate of any of the patient clusters;
• predict a progression rate for a new patient in accordance with the one patient cluster to which the new patient is assigned, and
• an output device operative to output the progression rate for the new patient.

A method for predicting Neurodegenerative Disease (ND) state performed on a computing device having a processor, memory, and one or more code sets stored in the memory and executed in the processor, the method comprising:

• receiving feature-based representation of ND patients, the feature-based representation including static and dynamic features;
• randomly assigning a plurality of patients among a plurality of patient clusters, each patient represented with static and dynamic features;
• training a long short-term memory (LSTM)-based classifier for each patient cluster, each respective classifier operative to predict a disease state for a plurality of patients of an associated patient cluster, the disease state characterized by Functionality Measure (FM) items;
• iteratively running a training cycle until a performance measure is achieved, the training cycle including:
• predicting a disease state for a plurality of patients with each of the respective classifiers,
• reassigning at least one patient to a patient cluster best matching his disease state, and
• retraining each of the respective classifiers using a plurality of patients now associated with each of the patient clusters;
• and
• predicting disease state of a new patient in accordance with the cluster to which the new patient is assigned.

A system for predicting Neurodegenerative Disease (ND) state and progression rate, the system comprising:

• an input device operative to receive feature-based representation of ND patients, the feature-based representation including static and dynamic features;
• a computing device configured to:
• randomly assign a plurality of patients among a plurality of patient clusters, each patient represented with static and dynamic features,
• train a long short-term memory (LSTM)-based classifier for each patient cluster, each respective classifier operative to predict a disease state for a plurality of patients of an associated patient cluster, the disease state characterized by Functionality Measure (FM) items;
• iteratively run a training cycle until a performance measure is achieved, the training cycle including:
• predicting a disease state for a plurality of patients with each of the respective classifiers,
• reassigning at least one patient to a patient cluster best matching his disease state, and
• retraining each of the respective classifiers using a plurality of patients now associated with each of the patient clusters;
• predict disease state of a new patient in accordance with the cluster to which the new patient is assigned;
• and
• an output device operative to output the disease state.

A method for predicting Neurodegenerative Disease (ND) progression rate performed on a computing device having a processor, memory, and one or more code sets stored in the memory and executed in the processor, the method comprising:

• receiving feature-based representation of ND patients, the feature based representation implemented as Functionality Measure (FM) items;
• calculating individual FM item progression rates;
• establishing a multi-dimensional patient representation based on the FM item progression rates;
• clustering the multi-dimensional patient representation into distinct patient clusters;
• identifying an optimal number of patient clusters from among patient data in accordance with data driven optimization scheme;
• and
• predicting a progression rate for a new patient in accordance with a patient cluster of the distinct patient clusters to which the new patient is assigned.

An integrated system for predicting Neurodegenerative Disease (ND) progression rate, the system comprising:

• an input device operative to receive feature-based patient representations;
• a computing device configured to:
• receive feature-based representation of ND patients, the feature based representation implemented as Functionality Measure (FM) items;
• calculate individual FM item progression rates;
• establish a multi-dimensional patient representation based on the FM item progression rates;
• cluster the multi-dimensional patient representation into distinct patient clusters;
• identify an optimal number of patient clusters from among patient data in accordance with a data driven optimization scheme;
• assign a new patient to one of the patient clusters in accordance with a best match between a progression rate of the new patient and a progression rate of any of the patient clusters;
• predict a progression rate for the new patient in accordance with the one patient cluster to which the new patient is assigned, and
• an output device operative to output the progression rate.