Next best action based on physical predictive model of patient physical health

Abstract

A method may include collecting sensor data related to a physical or mental health of a patient. The method may include determining a series of physical evidence nodes. The method may include determining a series of physical states of the patient. Each physical state is associated with a single physical evidence node. The method may include determining an expected value for each physical evidence node based on a physical baseline of the patient. The method may include determining, after each day, a value for each corresponding physical evidence node, wherein the value for each physical evidence node includes a value from each associated physical state. The method may include determining, after each day, a deviation of the patient from the expected value. The method may include generating a physical predictive model of the physical health of the patient based on the deviation of the patient from the expected value.

What is claimed is: 
     
         1 . A system to generate a physical predictive model of a physical health of a patient, the system comprising:
 one or more sensors configured to collect sensor data related to a physical or mental health of a patient that pertains to at least one of a diet pattern, a sleep pattern, an exercise pattern, an activity level, a heart rate, a posture, a stress, a blood pressure variation, a blood glucose, a heart rhythm, a smoking status, a pain level, and a GPS data;   a memory configured to store the sensor data; and   a processor coupled to the one or more sensors and the memory, wherein the processor is configured to perform or control performance of executable operations comprising:
 determining a series of physical evidence nodes, wherein the physical evidence nodes correspond with at least one of a schedule and one or more goals for the patient during a single day within a period of time; 
 determining a series of physical states of the patient, wherein each physical state is associated with a single physical evidence node; 
 determining an expected value for each physical evidence node based on a physical baseline of the patient; 
 determining, after each day, a value for each corresponding physical evidence node, wherein the value for each physical evidence node includes a value from each physical state associated with the physical evidence node; 
 determining, after each day, a deviation of the patient from the expected value for each corresponding physical evidence node; and 
 generating a physical predictive model of the physical health of the patient based on the deviation of the patient from the expected value for each corresponding physical evidence node. 
   
     
     
         2 . The system of  claim 1 , wherein generating the physical predictive model of the physical health of the patient based on the deviation of the patient from the expected value for each corresponding physical evidence node comprises determining a probability of the patient being in a particular physical state according to P(πt|E day1 , E day2 , . . . , E dayn ), and wherein E day1  through E dayn  represent the physical evidence nodes for each day of the period of time and π t  represents the particular physical state. 
     
     
         3 . The system of  claim 2 , wherein generating the physical predictive model of the physical health of the patient based on the deviation of the patient from the expected value for each corresponding physical evidence node further comprises marginalizing a probability of the patient being in a particular physical state on a particular day according to E Day i ⊥{π x : x≠i}, and wherein x represents a day number in a series of days, E dayi  represent the physical evidence node associated with the particular day, and π x  represents the particular physical state. 
     
     
         4 . The system of  claim 1 , wherein generating the physical predictive model of the physical health of the patient based on the deviation of the patient from the expected value for each corresponding physical evidence node comprises determining a significance level of a particular physical state according to:
   α t (π t )= P (π t   ,E   day 1:t )=Σ π     t−1     P (π t ,π t−1   ,E   day 1:t );
     α t (π t )= P ( E   day t |π t )Σ π     t−1     P ( E   Day t |π t ,π t−1   ,E   day 1:t−1 ) P (π t |π t−1   ,E   day 1:t−1 ) P (π t−1   ,E   day 1:t−1 );
     α t (π t )= P ( E   day t |π t )Σ π     t−1     P (π t |π t−1 )α t−1 (π t−1 );
   
       and wherein E day1:t  represent the physical evidence nodes for each day of the period of time, π t  represents the particular physical state, and α t  represents the significance level of the particular physical state. 
     
     
         5 . The system of  claim 4 , wherein generating the physical predictive model of the physical health of the patient based on the deviation of the patient from the expected value for each corresponding physical evidence node further comprises determining a probability of the patient being in a particular physical state on a particular day according to 
       
         
           
             
               
                 
                   P 
                   ⁡ 
                   ( 
                   
                     
                       π 
                       t 
                     
                     | 
                     
                       E 
                       
                         
                           day 
                           ⁢ 
                               
                           1 
                         
                         : 
                         t 
                       
                     
                   
                   ) 
                 
                 = 
                 
                   
                     α 
                     
                       t 
                       ⁡ 
                       ( 
                       
                         π 
                         t 
                       
                       ) 
                     
                   
                   
                     
                       ∑ 
                       
                         i 
                         ∈ 
                         Ω 
                       
                     
                     
                       α 
                       
                         t 
                         ⁡ 
                         ( 
                         
                           π 
                           t 
                         
                         ) 
                       
                     
                   
                 
               
               , 
             
           
         
       
       and Ω represents the series of physical states of the patient corresponding with the particular day. 
     
     
         6 . The system of  claim 1 , wherein generating the physical predictive model of the physical health of the patient based on the deviation of the patient from the expected value for each corresponding physical evidence node comprises determining a probability of the patient being in a particular physical state according to P(π t |E day1 , E day2 , . . . , E dayN )=[P πt *(Π n   i=1 P(E Day(i) |E Day(i−1) , E Day(i−2) , . . . , E Day1 , π t ))]/P(E Day1 , . . . , E DayN ), and wherein E day1  through E dayN  represent the physical evidence nodes for each day of the period of time and π t  represents the particular physical state. 
     
     
         7 . The system of  claim 1 , the system further comprising a chronic disease database and the executable operations further comprising determining a health risk score of the patient according to HS=f(SHC, CCB, LCC), wherein HS represents the health risk score, SHC represents a short term health score, CCB represents a chronic burden score, and LCC represents a lifestyle choice score, and wherein a quantification is based on a statistical distribution of the physical predictive model and of data included in the chronic disease database with respect to at least one of the CCB and the LCC of the patient. 
     
     
         8 . The system of  claim 1 , the system further comprising a questionnaire module configured to receive patient input in response to one or more questions directed to a quality of health of the patient, wherein the physical evidence nodes are further determined based on the patient input. 
     
     
         9 . The system of  claim 1 , wherein each day represents a period of time between a start node that relates to the patient waking up and an end node that relates to the patient falling asleep. 
     
     
         10 . The system of  claim 1 , the executable operations further comprising:
 determining a patient recovery rate pre-symptom based on the sensor data; and   determining a tiredness level post symptom based on the sensor data, wherein the physical baseline of the patient is based on the patient recovery rate pre-symptom and the tiredness level post symptom.   
     
     
         11 . The system of  claim 1 , wherein each physical state of the patient corresponds to one of a specific disease, a specific habit, and a specific risk of the patient. 
     
     
         12 . The system of  claim 11 , wherein each physical state of the patient is mapped to one of a different weather type and a season of a year. 
     
     
         13 . The system of  claim 1 , wherein the physical baseline of the patient is based on a distribution specific to a similar patient. 
     
     
         14 . A method to generate a physical predictive model of a physical health of a patient, the method comprising:
 collecting sensor data related to a physical or mental health of a patient that pertains to at least one of a diet pattern, a sleep pattern, an exercise pattern, an activity level, a heart rate, a posture, a stress, a blood pressure variation, a blood glucose, a heart rhythm, a smoking status, a pain level, and a GPS data;   determining a series of physical evidence nodes, wherein the physical evidence nodes correspond with at least one of a schedule and one or more goals for the patient during a single day within a period of time;   determining a series of physical states of the patient, wherein each physical state is associated with a single physical evidence node;   determining an expected value for each physical evidence node based on a physical baseline of the patient;   determining, after each day, a value for each corresponding physical evidence node, wherein the value for each physical evidence node includes a value from each physical state associated with the physical evidence node;   determining, after each day, a deviation of the patient from the expected value for each corresponding physical evidence node; and   generating a physical predictive model of the physical health of the patient based on the deviation of the patient from the expected value for each corresponding physical evidence node.   
     
     
         15 . The method of  claim 14 , wherein generating the physical predictive model of the physical health of the patient based on the deviation of the patient from the expected value for each corresponding physical evidence node comprises determining a probability of the patient being in a particular physical state according to P(πt|E day1 , E day2 , . . . , E dayn ), and wherein E day1  through E dayn  represent the physical evidence nodes for each day of the period of time and π t  represents the particular physical state. 
     
     
         16 . The method of  claim 15 , wherein generating the physical predictive model of the physical health of the patient based on the deviation of the patient from the expected value for each corresponding physical evidence node further comprises marginalizing a probability of the patient being in a particular physical state on a particular day according to E Day i   {π x : x≠i}, and wherein x represents a day number in a series of days, E dayi  represent the physical evidence node associated with the particular day, and π x  represents the particular physical state. 
     
     
         17 . The method of  claim 14 , wherein generating a physical predictive model of a physical health of a patient based on the deviation of the patient from the expected value for each corresponding physical evidence node comprises determining a significance level of a particular physical state according to:
   α t (π t )= P (π t   ,E   day 1:t )=Σ π     t−1     P (π t ,π t−1   ,E   day 1:t );
     α t (π t )= P ( E   day t |π t )Σ π     t−1     P ( E   Day t |π t ,π t−1   ,E   day 1:t−1 ) P (π t |π t−1   ,E   day 1:t−1 ) P (π t−1   ,E   day 1:t−1 );
     and     α t (π t )= P ( E   day t |π t )Σ π     t−1     P (π t |π t−1 )α t−1 (π t−1 );
   
       and wherein E day1:t  represent the physical evidence nodes for each day of the period of time, π t  represents the particular physical state, and α t  represents the significance level of the particular physical state. 
     
     
         18 . The method of  claim 14 , the method further comprising:
 receiving chronic disease data; and   determining a health risk score of the patient according to HS=f(SHC, CCB, LCC), wherein HS represents the health risk score, SHC represents a short term health score, CCB represents a chronic burden score, and LCC represents a lifestyle choice score, and wherein a quantification is based on a statistical distribution of the physical predictive model and of the chronic disease data with respect to at least one of the CCB and the LCC of the patient.   
     
     
         19 . The method of  claim 14 , the method further comprising receiving patient input in response to one or more questions directed to a quality of health of the patient, wherein the physical evidence nodes are further determined based on the patient input. 
     
     
         20 . The method of  claim 14 , the method further comprising:
 determining a patient recovery rate pre-symptom based on the sensor data; and   determining a tiredness level post symptom based on the sensor data, wherein the physical baseline of the patient is based on the patient recovery rate pre-symptom and the tiredness level post symptom.