Clinical support systems and methods

A computer implemented method for clinical performance measurement of a clinician treating a patient subject, said method comprising:

• defining a range of an acceptable treatment variable values for a treatment variable of the patient subject;
• the treatment variable is an analyte level of the patient subject's blood and the acceptable treatment variable values is a range of the analyte level;
• the clinician selecting a selected intervention for the patient subject;
• the clinician performing the selected intervention on the patient subject;
• the selected intervention comprising a medicine change for the patient subject;
• determining an affected treatment variable value for the treatment variable of the patient subject from a sensor after performing the selected intervention;
• a processor based clinical support system modeling the selected intervention on the treatment variable over a predicted time with a predictive model;
• the predictive model trained with a plurality of affected treatment variable values of the patient subject;
• determining, with the predictive model, a plurality of predicted affected treatment variable values over the predicted time;
• the processor based clinical support system automatically determining a selection performance measure as the clinical performance measurement of the clinician for the selected intervention by comparing the affected treatment variable value and the plurality of predicted affected treatment variable values to the acceptable treatment variable values;
• performing a plurality of selected interventions on the patient subject over a performance time period;
• determining a plurality of the selection performance measures over the performance time period;
• each of the plurality of the selection performance measures associated with one or more time in the performance time period;
• determining whether one or more of the plurality of the selection performance measures is one of a good selection performance measure or a bad selection performance measure wherein the good selection performance measure is correlated with one of the predicted affected treatment variable values being within the acceptable treatment variable values and the bad selection performance measure is correlated with one of the predicted affected treatment variable values being outside of the acceptable treatment variable values;
• identifying a performance indicator by performing a first pattern recognition analysis on the plurality of selected interventions and the plurality of the selection performance measures over the performance time period whereby the performance indicator identifies an association of one or more of the plurality of selected interventions with one or more of the good selection performance measure or the bad selection performance measure;
• the first pattern recognition analysis comprises a stochastic search algorithm configured to compare and find patterns between the plurality of the selection performance measures and the treatment variable value comprises a predicted future analyte level existent in a historical dataset of treatment variable values;
• the clinician performing the plurality of selected interventions on the patient subject over a treatment time period;
• determining the plurality of affected treatment variable values over the treatment time period;
• identifying the one or more time wherein the affected treatment variable value is one of a good affected treatment variable value or a bad affected treatment variable value;
• identifying a treatment indicator by performing a second pattern recognition analysis on the plurality of selected interventions on the patient subject and the plurality of affected treatment variable values at the one or more time over the treatment time period whereby the treatment indicator identifies an association of one or more of the plurality of selected interventions with one or more of the good affected treatment variable value or the bad affected treatment variable value;
• and
• the second pattern recognition analysis comprises a second stochastic search algorithm configured to find patterns of the plurality of selected interventions on the patient subject to compare to a plurality of historical selected interventions from a historical dataset of selected interventions and the treatment variable values comprises a predicted future analyte level value existent in the historical dataset of treatment variable values.