Method for predicting risk of delirium and device for predicting risk of delirium using the same

1. A delirium risk predicting method implemented by a processor, the method comprising:

receiving blood data, severity evaluation data, mental state evaluation data, and bio signal data, medication data, and medical treatment data, for a subject, wherein the blood data is taken from a blood sample separated from the subject and includes one of blood urea nitrogen, pH, HCO3, albumin (Alb), hemoglobin, hematocrit (Hct), Bilirubin (BILI), Na, and neutrophil to lymphocyte ratio (NLR);

predicting a delirium risk for the subject, using a delirium risk prediction model configured to predict a delirium risk, based on the blood data, the severity evaluation data, the mental state evaluation data, the bio signal data, the medication data, and the medical treatment data; and

providing the delirium risk predicted for the subject,

wherein the blood data, the medication data, and the medical treatment data are defined as initial data which is evaluated or measured only once for the subject,

wherein the delirium risk prediction model is a model trained by receiving learning data configured by at least one of blood data, severity evaluation data, mental state evaluation data, and bio signal data, medication data, and medical treatment data for a delirium sample subject and a normal sample subject;

calculating a relevance score to the delirium for the learning data; and

determining delirium related learning data within a predetermined ranking, based on the relevance score and predicting to be delirium or normal based on delirium related learning data,

wherein the relevance score is defined as a relevance between an input value and an output value indicating to be delirium or normal state.

2. The delirium risk predicting method according to claim 1 , wherein the mental state evaluation data is a Richmond agitation and sedation scale (RASS) or state-trait anxiety inventory (STAI) score.

3. The delirium risk predicting method according to claim 1 , wherein the severity evaluation data includes acute physiology and chronic health evaluation (APACHE II) score.

4. The delirium risk predicting method according to claim 1 , wherein the bio signal data includes at least one of a pulse rate, a respiration rate, a body temperature, a systolic blood pressure (SBP), and a diastolic blood pressure (DBP).

5. The delirium risk predicting method according to claim 1 , wherein the medication data includes at least one of Ultracet™, Midazolam™, Ultiva™, Ativan™, Pofol™, fentanyl, Precedex™, IR codon™, TARGIN 20/10™, Peridol™, Risperdal™, Zyprexa™, Seroquel™, Abilify™, pethidine, Durogesic patch™, morphine, and Mypol™.

6. The delirium risk predicting method according to claim 1 , wherein the medical treatment data includes at least one medical treatment data of catheter, Foley, mechanical ventilator, restraint, and drainage.

7. The delirium risk predicting method according to claim 1 , wherein the initial data is defined as initial data which is initially evaluated or measured for the subject in a predetermined time unit for a hospitalization period.

8. The delirium risk predicting method according to claim 1 , further comprising:

determining a delirium inducing drug based on the medication data,

wherein the providing of the delirium risk includes:

providing the delirium risk predicted for the subject and the delirium inducing drug.

9. The delirium risk predicting method according to claim 1 , wherein the providing of the delirium risk includes providing a notification of a delirium risk for the subject when the delirium risk for the subject is predicted by the delirium risk prediction model.

10. The delirium risk predicting method according to claim 1 , wherein the calculating a relevance score to the delirium includes:

calculating a relevance score to the delirium for the learning data, using a layer-wise relevance propagation (LRP) algorithm and

wherein determining delirium related learning data includes:

determining delirium related learning data within a predetermined ranking, based on the relevance score.

11. A delirium risk predicting method implemented by a processor, the method comprising:

receiving blood data, severity evaluation data, mental state evaluation data, bio signal data, medication data, and medical treatment data, for a subject, wherein the blood data is taken from a blood sample separated from the subject and includes one of blood urea nitrogen, pH, HCO3, albumin (Alb), hemoglobin, hematocrit (Hct), Bilirubin (BILI), Na, and neutrophil to lymphocyte ratio (NLR);

predicting a delirium risk for the subject, using a delirium risk prediction model configured to predict a delirium risk, based on the blood data, the severity evaluation data, the mental state evaluation data, the bio signal data, the medication data, and the medical treatment data; and

providing the delirium risk predicted for the subject,

wherein the blood data, the medication data, and the medical treatment data are defined as initial data which is evaluated or measured only once for the subject, and,

wherein the delirium risk prediction model is a multi-layer perceptron (MLP) algorithm based prediction model, and a model trained by receiving learning data configured by at least one of blood data, severity evaluation data, mental state evaluation data, bio signal data, medication data, and medical treatment data for a delirium sample subject and a normal sample subject;

calculating a relevance score to the delirium for the learning data; and

determining delirium related learning data within a predetermined ranking, based on the relevance score and predicting to be delirium or normal based on the delirium related learning data,

wherein the relevance score is defined as a relevance between an input value and an output value indicating to be delirium or normal state.

12. A delirium risk predicting device implemented by a processor, the device comprising:

a receiver configured to receive blood data, severity evaluation data, mental state evaluation data, bio signal data, medication data, and medical treatment data, for a subject, wherein the blood data is taken from a blood sample separated from the subject and includes one of blood urea nitrogen, pH, HCO3, albumin (Alb), hemoglobin, hematocrit (Hct), Bilirubin (BILI), Na, and neutrophil to lymphocyte ratio (NLR); and

a processor configured to communicate with the receiver,

wherein the processor is further configured to predict a delirium risk for the subject, using a delirium risk prediction model configured to predict a delirium risk, based on the blood data, the severity evaluation data, the mental state evaluation data, the bio signal data, the medication data, and the medical treatment data and provide the delirium risk predicted for the subject, and the blood data, the medication data, and the medical treatment data are defined as initial data which is evaluated or measured only once for the subject,

wherein the delirium risk prediction model is a model trained by receiving learning data configured by at least one of blood data, severity evaluation data, mental state evaluation data, and bio signal data, medication data, and medical treatment data for a delirium sample subject and a normal sample subject;

wherein the processor is further configured to calculate a relevance score to the delirium for the learning data and to determine delirium related learning data within a predetermined ranking, based on the relevance score and predicting to be delirium or normal based on delirium related learning data,

wherein the relevance score is defined as a relevance between an input value and an output value indicating to be delirium or normal state.

13. The delirium risk predicting device according to claim 12 , wherein the initial data is defined as initial data which is initially evaluated or measured for the subject in a predetermined time unit for a hospitalization period.

14. The delirium risk predicting device according to claim 12 , wherein the processor is further configured to determine a delirium inducing drug based on the medication data and provide the delirium risk predicted for the subject and the delirium inducing drug.

15. The delirium risk predicting device according to claim 12 , wherein the processor is further configured to provide a notification of a delirium risk for the subject when the delirium risk for the subject is predicted by the delirium risk prediction model.

16. The delirium risk predicting device according to claim 12 , wherein the medication data includes at least one of Ultracet™, Midazolam™, Ultiva™, Ativan™, Pofol™, fentanyl, Precedex™, IR codon™, TARGIN 20/10™, Peridol™, Risperdal™, Zyprexa™, Seroquel™, Abilify™, pethidine, Durogesic patch™, morphine, and Mypol™.

17. The delirium risk predicting device according to claim 12 , wherein the medical treatment data includes at least one medical treatment data of catheter, Foley, mechanical ventilator, restraint, and drainage.