METHODS AND SYSTEMS FOR PREDICTING A HEALTH CONDITION OF A HUMAN SUBJECT
Disclosed are embodiments of methods and systems for predicting a health condition of a first human subject. The method comprises extracting a historical data including physiological parameters of one or more second human subjects. A latent variable is determined based on an inverse cumulative distribution of a transformed historical data, determined by ranking of the historical data. Further, one or more parameters of a first distribution, deterministic of health conditions in the historical data, are determined based on the latent variable. For each physiological parameter, a random variable is sampled from a second distribution of the physiological parameter based on the one or more parameters. Further, based on the random variable, the latent variable is updated. Thereafter, the one or more parameters are re-estimated based on the updated latent variable. Based on the first distribution a classifier is trained to predict the health condition of the first human subject.
1 . A method for predicting a health condition of a first human subject, the method comprising:
extracting, by one or more processors, a historical data comprising a measure of one or more physiological parameters associated with each of one or more second human subjects;
determining, by said one or more processors, a latent variable based on an inverse cumulative distribution of a transformed historical data, wherein said transformed historical data is determined by ranking of said historical data;
estimating, by said one or more processors, one or more parameters of a first distribution deterministic of one or more health conditions in said historical data, based on said latent variable;
for each physiological parameter from said one or more physiological parameters:
sampling, by said one or more processors, a random variable from a second distribution of said physiological parameter, based on said one or more parameters;
updating, by said one or more processors, said latent variable based on said random variable;
re-estimating, by said one or more processors, said one or more parameters based on said updated latent variable;
training, by said one or more processors, a classifier based on said first distribution;
receiving, by said one or more processors, a measure of said one or more physiological parameters associated with said first human subject; and
predicting, by said one or more processors, said health condition of said first human subject by utilizing said classifier based on said received measure of said one or more physiological parameters associated with said first human subject.
2 . The method of claim 1 , wherein said one or more physiological parameters comprise at least one of a blood glucose level, a blood pressure, an age, a cholesterol level, a heart rate, a breath carbon-dioxide concentration, or a breath oxygen concentration.
3 . The method of claim 1 , wherein said one or more parameters are estimated by utilizing one of a Gibbs sampling technique or an Expectation-Maximization (EM) technique.
4 . The method of claim 1 , wherein said second distribution is truncated based on a lower bound and an upper bound of said latent variable for said physiological parameter.
5 . The method of claim 1 , wherein said first distribution corresponds to a Gaussian Copula Mixture distribution.
6 . The method of claim 1 , wherein said one or more parameters comprise at least one of a mean, a covariance matrix, and a mixing proportion, of a cluster component associated with said first distribution.
7 . The method of claim 1 further comprising determining, by said one or more processors, said second distribution of said physiological parameter for each of said one or more health conditions.
8 . The method of claim 1 , wherein said updating of said latent variable is based on a mixing proportion of each of said one or more health conditions in said historical data.
9 . The method of claim 1 , wherein said ranking of said historical data corresponds to an extended rank likelihood.
10 . The method of claim 1 , wherein a data type associated with said historical data corresponds to at least one of a numerical data type or a categorical data type.
11 . The method of claim 1 , wherein said historical data corresponds to a multivariate dataset from which said one or more health conditions are identifiable based on said first distribution.
12 . The method of claim 1 , wherein each of said one or more health conditions corresponds to at least one of a disease risk, a disease symptom, an onset of a disease, a recovery from a disease, or an effect of medications for a disease.
13 . A system for predicting a health condition of a first human subject, the system comprising:
one or more processors configured to:
extract a historical data comprising a measure of one or more physiological parameters associated with each of one or more second human subjects;
determine a latent variable based on an inverse cumulative distribution of a transformed historical data, wherein said transformed historical data is determined by ranking of said historical data;
estimate one or more parameters of a first distribution deterministic of one or more health conditions in said historical data, based on said latent variable;
for each physiological parameter from said one or more physiological parameters:
sample a random variable from a second distribution of said physiological parameter, based on said one or more parameters;
update said latent variable based on said random variable;
re-estimate said one or more parameters based on said updated latent variable;
train a classifier based on said first distribution;
receive a measure of said one or more physiological parameters associated with said first human subject; and
predict said health condition of said first human subject by utilizing said classifier based on said received measure of said one or more physiological parameters associated with said first human subject.
14 . The system of claim 13 , wherein said one or more parameters are estimated by utilizing one of a Gibbs sampling technique or an Expectation-Maximization (EM) technique.
15 . The system of claim 13 , wherein said second distribution is truncated based on a lower bound and an upper bound of said latent variable for said physiological parameter.
16 . The system of claim 13 , wherein said first distribution corresponds to a Gaussian Copula Mixture distribution.
17 . The system of claim 13 , wherein said updating of said latent variable is based on a mixing proportion of each of said one or more health conditions in said historical data.
18 . The system of claim 13 , wherein said ranking of said historical data corresponds to an extended rank likelihood.
19 . A computer program product for use with a computing device, the computer program product comprising a non-transitory computer readable medium, wherein the non-transitory computer readable medium stores a computer program code for predicting a health condition of a first human subject, wherein the computer program code is executable by one or more processors in the computing device to:
extract a historical data comprising a measure of one or more physiological parameters associated with each of one or more second human subjects;
determine a latent variable based on an inverse cumulative distribution of a transformed historical data, wherein said transformed historical data is determined by ranking of said historical data;
estimate one or more parameters of a first distribution deterministic of one or more health conditions in said historical data, based on said latent variable;
for each physiological parameter from said one or more physiological parameters:
sample a random variable from a second distribution of said physiological parameter, based on said one or more parameters;
update said latent variable based on said random variable;
re-estimate said one or more parameters based on said updated latent variable; and
train a classifier based on said first distribution;
receive a measure of said one or more physiological parameters associated with said first human subject; and
predict said health condition of said first human subject by utilizing said classifier based on said received measure of said one or more physiological parameters associated with said first human subject.