| 主权项 |
1. One or more computer storage media storing computer-useable instructions that, when used by one or more computing devices, cause the one or more computing devices to perform a method comprising:
(a) receiving training data associated with a plurality of human infant cases, the plurality of human infant cases having been selected based on inclusion-exclusion criteria, the training data including data for a plurality of variables; (b) partitioning the training data into a training data subset and a test data subset, each of the training data subset and the test data subset including training data associated with a portion of the plurality of human infant cases; (c) employing a machine-learning technique to generate a classifier based on the training data subset; (d) assessing whether the classifier satisfies a predetermined receiver operating characteristic (ROC) c-statistic; (e) if the classifier does not satisfy the predetermined ROC c-statistic, repeating (c) and (d) until a classifier is generated that satisfies the predetermined ROC; (f) if the classifier satisfies the predetermined ROC c-statistic, validation testing the classifier using the test data subset; (g) assessing whether an optimum minimum has been achieved based on output from the validation testing; (h) if an optimum minimum has not been achieved, repeating (c)-(g) until an optimum minimum has been achieved; (i) if an optimum minimum has been achieved, generating a multi-variable logistic regression model capable of calculating a probability of clinically significant hyperbilirubinemia using the plurality of variables; (j) assessing whether statistical performance of the multi-variable logistic regression model satisfies a predetermined requirement; (k) if the statistical performance of the multi-variable logistic regression model does not satisfy the predetermined requirement, repeating (c)-(j) until a multi-variable logistic regression model is generated that satisfies the predetermine requirement; (l) receiving input data for a human infant, the input data including laboratory test results for the human infant determined from measurements received at a single measurement-session time and including a time from birth associated with the laboratory test results; (m) calculating a probability of clinically significant hyperbilirubinemia using the input data for the human infant and the multi-variable logistic regression model; (n) comparing the calculated probability to one or more thresholds for hyperbilirubinemia intervention to determine one or more risk levels; and (o) communicating for presentation to a clinician the one or more risk levels. |
