Forecasting a patient vital measurement for healthcare analytics
What is disclosed is a system and method for forecasting and imputing an unknown vital measurement of a patient. Temporally successive patient vital measurements are received which comprise irregularly sampled observations {y 1 , . . . , y N }, where y j denotes the j th observation at time t j , and N is the number of samples. The vital measurements are then provided to a model trained using historical data of patient vital measurements. The model generates a parameter set θ=(A,B,C), where A is a state transition matrix, B is a control matrix, and C is a matrix which maps state-space variables to observation variables. The parameters are used to obtain state-space variable z t which, in turn, is used to forecast an unknown observation y N+1 or to impute an unknown observation y t , where 1<t<N. The historical data is then updated with the forecasted observation y N+1 or the imputed unknown observation y t .
1. A computer implemented method for obtaining an unknown patient vital measurement for healthcare analytics, the method comprising:
receiving, by a processor, temporally successive patient vital measurements comprising irregularly sampled observations {y 1 , . . . , y N }, where y j denotes the j th observation at time t j , and N is the number of samples;
communicating, by a processor, the temporally successive vital measurements to a model trained using historical data of patient vital measurements, the model generating a parameter set θ=(A,B,C,Q,R), where A is a state transition matrix, B is a control matrix, C is a matrix which maps state-space variables to observation variables, Q is an amount of noise in the state-space variables, and R is an amount of noise in the observation variables;
calculating, by a processor, state-space variables z t using the parameter set θ generated by the model; and
performing, by a processor, one of: forecasting an unknown observation y N+1 , and imputing an unknown observation y t , where 1<t<N.
2. The computer implemented method of claim 1 , wherein the patient vital measurements are obtained from Electronic Medical Records.
3. The computer implemented method of claim 1 , wherein the vital measurements are any of: blood pressure, respiration rate, heart rate, body temperature, blood glucose, oxygen saturation, blood CO 2 , Glasgow Coma Scale, and a combination hereof.
4. The computer implemented method of claim 1 , wherein z t comprises:
z t =Az t−1 +BΔ t,t-1 +∈ t
where Δ t,t-1 denotes a time difference between the t th and the (t−1) th observation, ∈ t ˜ (0, Q) is approximated by a zero-mean Gaussian with an unknown co-variance, and Q is an amount of noise in the state-space variables.
5. The computer implemented method of claim 1 , wherein y t comprises:
y t =Cz t +δ t
where δ t ˜ (0, R) is approximated by a zero-mean Gaussian with an unknown co-variance and R is an amount of noise in the observation variables.
6. The computer implemented method of claim 1 , further comprising communicating the obtained observation to an ICU Admission Prediction System to help identify patients requiring ICU admission.
7. The computer implemented method of claim 1 , further comprising communicating the obtained observation to an Emerging Complications Prediction System to help identify patients at risk for developing complications during their hospital stay.
8. The computer implemented method of claim 1 , further comprising updating the historical data of patient vital measurements with the forecasted observation y N+1 .
9. The computer implemented method of claim 1 , further comprising updating the historical data of patient vital measurements with the imputed observation y t .
10. A system for obtaining an unknown patient vital measurement for healthcare analytics, the system comprising:
a storage device; and
a processor in communication with the storage device, the processor executing machine readable instructions for performing:
receiving temporally successive patient vital measurements comprising irregularly sampled observations {y 1 , . . . , y N }, where y j denotes the j th observation at time t j , and N is the number of samples;
communicating the temporally successive vital measurements to a model trained using historical data of patient vital measurements, the model generating a parameter set θ=(A,B,C,Q,R), where A is a state transition matrix, B is a control matrix, C is a matrix which maps state-space variables to observation variables, Q is an amount of noise in the state-space variables, and R is an amount of noise in the observation variables;
calculating state-space variables z t using the parameter set θ generated by the model;
performing one of: forecasting an unknown observation y N+1 , and imputing an unknown observation y t , where 1<t<N; and
storing the obtained observation to the storage device.
11. The system of claim 10 , wherein the patient vital measurements are obtained from Electronic Medical Records.
12. The system of claim 10 , wherein z t comprises:
z t =Az t−1 +BΔ t,t-1 +∈ t
where Δ t,t-1 denotes a time difference between the t th and the (t−1) th observation, ∈ t ˜ (0, Q) is approximated by a zero-mean Gaussian with an unknown co-variance, and Q is an amount of noise in the state-space variables.
13. The system of claim 10 , wherein y t comprises:
y t =Cz t +δ t
where δ t ˜ (0, R) is approximated by a zero-mean Gaussian with an unknown co-variance and R is an amount of noise in the observation variables.
14. The system of claim 10 , further comprising communicating the obtained observation to an ICU Admission Prediction System to help identify patients requiring ICU admission.
15. The system of claim 10 , further comprising communicating the obtained observation to an Emerging Complications Prediction System to help identify patients at risk for developing complications during their hospital stay.
16. The system of claim 10 , further comprising updating the historical data of patient vital measurements with the forecasted observation y N+1 .
17. The system of claim 9 , further comprising updating the historical data of patient vital measurements with the imputed observation y t .
18. The system of claim 10 , wherein the vital measurements are any of: blood pressure, respiration rate, heart rate, body temperature, blood glucose, oxygen saturation, blood CO 2 , Glasgow Coma Scale, and a combination hereof.