Systems and methods for enhanced outlier detection

1 . A computer-implemented method, comprising:

receiving, by one or more computing devices, a first claim including a provider, an illness, a treatment, a code, and a risk score, wherein the risk score is based on the illness and the treatment;

in real-time in response to receiving the first claim:

grouping, by the one or more computing devices, based on the code, the provider into a specialty group;

updating, by the one or more computing devices, a metric associated with the provider based on the first claim;

computing, by the one or more computing devices and based on historical risk score and metric data of the specialty group corresponding to a dynamic rolling time window, a risk impact coefficient, wherein the risk impact coefficient represents a linear relationship between the metric and the risk score;

determining, by the one or more computing devices, an impact of the risk score based on the risk impact coefficient and the risk score;

subtracting, by the one or more computing devices, the impact of the risk score from the metric to generate a normalized metric value;

determining, by the one or more computing devices, that the normalized metric value is an outlier by:

comparing, by the one or more computing devices, the normalized metric value to a threshold corresponding to the specialty group; and

determining, by the one or more computing devices and based on the specialty group, that the normalized metric value is associated with a statistically significant number of claims; and

in response to determining that the normalized metric value is an outlier,

transmitting, by the one or more computing devices, an alert to a client device associated with a user, wherein the alert is configured to cause a display of the client device to display an indication of the outlier.

2 . The computer-implemented method of claim 1 , wherein comparing the normalized metric value to the threshold comprises:

generating, by the one or more computing devices, a first quartile value and a second quartile value based on a plurality of normalized metric values corresponding to a plurality of providers of the specialty group;

determining, by the one or more computing devices, an interquartile range based on the first quartile value and the second quartile value;

calculating, by the one or more computing devices, a median value of the interquartile range;

calculating, by the one or more computing devices, a lower limit of the threshold based on the median value, a threshold multiplier, and the interquartile range;

calculating, by the one or more computing devices, an upper limit of the threshold based on the median value, the threshold multiplier, and the interquartile range; and

determining, by the one or more computing devices, the normalized metric value is an outlier based on the normalized metric value being greater than the upper limit of the threshold or less than the lower limit of the threshold.

3 . The method of claim 2 , further comprising:

determining, by the one or more computing devices, one of the lower limit of the threshold or the upper limit of the threshold is less than a degradation value; and

in response to the determining, increasing, by the one or more computing devices, one of the lower limit of the threshold or the upper limit of the threshold by a degradation factor.

4 . The computer-implemented method of 1 , further comprising:

updating, by the one or more computing devices, a permission associated with a medical device, wherein the updated permission is configured to prevent access by the provider.

5 . The computer-implemented method of claim 1 , wherein grouping the provider into the specialty group further comprises:

retrieving, by the one or more computing devices, the code from the first claim;

transmitting, by the one or more computing devices, a request for the specialty group to a data provider system, the request including the code and the provider;

receiving, by the one or more computing devices, the specialty group from the data provider system; and

assigning, by the one or more computing devices, the provider to the specialty group.

6 . The computer-implemented method of claim 1 , wherein the alert includes a link to access a graphical user interface providing additional information associated with the alert.

7 . The computer-implemented method of claim 1 , wherein the metric is one of mortality rate, length of stay, readmission rate, audit rate, average total charges value, denial rate, or medicine prescription rate.

8 . The computer-implemented method of claim 1 , wherein determining that the normalized metric value is an outlier further comprises:

training a machine learning model on a plurality of previously received claims, wherein each claim of the plurality of previously received claims includes a respective first metric that is labeled as an outlier, and wherein the training comprises:

for each of the previously received claims, determining, by the machine learning model, a respective second metric correlated to the respective first metric of the previously received claim and labeling the respective second metric as an outlier; and

retraining the machine learning model on the plurality of previously received claims with the labeled respective second metrics;

wherein the training configures the machine learning model to determine whether a third metric is an outlier based on a claim associated with the third metric;

inputting the first claim into the machine learning model; and

determining, by the machine learning model, the normalized metric value associated with the provider is an outlier.

9 . The computer-implemented method of claim 1 , further comprising:

identifying, by the one or more computing devices, that the metric is included in an internal database; and

based on identifying that the metric is included in the internal database:

determining, by the one or more computing devices, a plurality of contributing claims associated with the provider, wherein the plurality of contributing claims contributed to the metric associated with the provider and have non-zero metric values;

determining, by the one or more computing devices, that the outlier is a false positive based on a number of the plurality of contributing claims being less than or equal to a coincidence threshold; and

in response to determining that the outlier is a false positive, suppressing, by the one or more computing devices, the alert.

10 . The computer-implemented method of claim 1 , wherein determining that the normalized metric value is associated with the statistically significant number of claims comprises:

determining, by the one or more computing devices, a number of eligible claims associated with the provider exceeds a minimum claims threshold.

11 . A system, comprising:

a memory; and

at least one processor coupled to the memory and configured to:

receive a first claim including a provider, an illness, a treatment, a code, and a risk score, wherein the risk score is based on the illness and the treatment;

in real-time in response to receiving the first claim:

group, based on the code, the provider into a specialty group;

update a metric associated with the provider based on the first claim;

compute, based on historical risk score and metric data of the specialty group corresponding to a dynamic rolling time window, a risk impact coefficient, wherein the risk impact coefficient represents a linear relationship between the metric and the risk score;

determine an impact of the risk score based on the risk impact coefficient and the risk score;

subtract the impact of the risk score from the metric to generate a normalized metric value;

determine that the normalized metric is an outlier, wherein the at least one processor is further configured to:

compare the normalized metric to a threshold corresponding to the specialty group; and

determine, based on the specialty group, that the normalized metric is associated with a statistically significant number of claims; and

in response to determining that the normalized metric is an outlier, transmit an alert to a client device associated with a user, wherein the alert is configured to cause a display of the client device to display an indication of the outlier.

12 . The system of claim 11 , wherein to compare the normalized metric value to the threshold, the at least one processor is further configured to:

generate a first quartile value and a second quartile value based on a plurality of normalized metric values corresponding to a plurality of providers of the specialty group;

determine an interquartile range based on the first quartile value and the second quartile value;

calculate a median value of the interquartile range;

calculate a lower limit of the threshold based on the median value, a threshold multiplier, and the interquartile range;

calculate an upper limit of the threshold based on the median value, the threshold multiplier, and the interquartile range; and

determine the normalized metric value is an outlier based on the normalized metric value being greater than the upper limit of the threshold or less than the lower limit of the threshold.

13 . The system of claim 12 , where in the at least one processor is further configured to:

determine one of the lower limit of the threshold or the upper limit of the threshold is less than a degradation value; and

in response to the determining, increase one of the lower limit of the threshold or the upper limit of the threshold by a degradation factor.

14 . The system of claim 11 , wherein the at least one processor is further configured to:

update a permission associated with a medical device, wherein the updated permission is configured to prevent access by the provider.

15 . The system of claim 11 , wherein the at least one processor is further configured to:

retrieve the code from the first claim;

transmit a request for the specialty group to a data provider system, the request including the code and the provider;

receive the specialty group from the data provider system; and

assign the provider to the specialty group.

16 . The system of claim 11 , wherein the alert includes a link to access a graphical user interface providing additional information associated with the alert.

17 . The system of claim 11 , wherein the metric is one of mortality rate, length of stay, readmission rate, audit rate, average total charges value, denial rate, or medicine prescription rate.

18 . A non-transitory computer-readable device having instructions stored thereon that, when executed by at least one computing device, cause the at least one computing device to perform operations comprising:

receiving a first claim including a provider, an illness, a treatment, a code, and a risk score, wherein the risk score is based on the illness and the treatment;

in real-time in response to receiving the first claim:

grouping, based on the code, the provider into a specialty group;

updating a metric associated with the provider based on the first claim;

computing, based on historical risk score and metric data of the specialty group corresponding to a dynamic rolling time window, a risk impact coefficient, wherein the risk impact coefficient represents a linear relationship between the metric and the risk score;

determining an impact of the risk score based on the risk impact coefficient and the risk score;

subtracting the impact of the risk score from the metric to generate a normalized metric value;

determining that the normalized metric value is an outlier by:

comparing the normalized metric value to a threshold corresponding to the specialty group; and

determining, based on the specialty group, that the normalized metric value is associated with a statistically significant number of claims; and

in response to determining that the normalized metric value is an outlier, transmitting an alert to a client device associated with a user, wherein the alert is configured to cause a display of the client device to display an indication of the outlier.

19 . The non-transitory computer-readable device of claim 18 , wherein when comparing the normalized metric value to the threshold, the operations further comprise:

generating a first quartile value and a second quartile value based on a plurality of normalized metric values corresponding to a plurality of providers of the specialty group;

determining an interquartile range based on the first quartile value and the second quartile value;

calculating a median value of the interquartile range;

calculating a lower limit of the threshold based on the median value, a threshold multiplier, and the interquartile range;

calculating an upper limit of the threshold based on the median value, the threshold multiplier, and the interquartile range; and

determining the normalized metric value is an outlier based on the normalized metric value being greater than the upper limit of the threshold or less than the lower limit of the threshold.

20 . The non-transitory computer-readable device of claim 19 , the operations further comprising:

determining one of the lower limit of the threshold or the upper limit of the threshold is less than a degradation value; and

in response to the determining, increasing one of the lower limit of the threshold or the upper limit of the threshold by a degradation factor.