Anomaly detection and performance assessment using a provider performance index

1 . A computer-implemented method, comprising:

selecting a first target, wherein the first target includes a first target value;

in real-time and based on selecting the first target:

retrieving a first dataset from a plurality of data sources, the first dataset comprising one or more features associated with the first target;

generating a first external weight by inputting the first target value and the one or more features associated with the first target into a first machine learning model corresponding to the first target, wherein the first machine learning model is configured to output an impact of the one or more features on the first target value;

generating a first adjusted target value by combining the first target value with the first external weight;

determining a first error value based on the first target value and the first adjusted target value; and

normalizing the first error value via a statistical measure;

selecting a second target, wherein the second target includes a second target value;

in real-time and based on selecting the second target:

retrieving a second dataset from the plurality of data sources, the second dataset comprising one or more features associated with the second target;

generating a second external weight by inputting the second target value and the one or more features associated with the second target into a second machine learning model corresponding to the second target, wherein the second machine learning model is configured to output an impact of the one or more features on the second target value;

generating a second adjusted target value by combining the second target value with the second external weight;

determining a second error value based on the second target value and the second adjusted target value; and

normalizing the second error value via the statistical measure; and

in response to normalizing the second error value or normalizing the first error value:

generating an index value based on the first normalized error value and the second normalized error value; and

transmitting, based on the index value being within a threshold percentile of a plurality of index values, an alert to one or more user computing devices.

2 . The computer-implemented method of claim 1 , wherein the index value is generated based on the average of the first normalized error value and the second normalized error value.

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

in response to generating the index value, automatically generating a message including the index value, the first target, and the second target; and

transmitting the message via a network to the one or more user computing devices.

4 . The computer-implemented method of claim 3 , wherein the message comprises a link to access a graphical user interface (GUI) via the network, wherein the link is unique to the generated index value.

5 . The computer-implemented method of claim 1 , wherein the statistical measure is a z-score.

6 . The computer-implemented method of claim 1 , wherein the first and second machine learning models include one or more weights corresponding to average age, percentage male, percentage female, continuously enrolled months, concurrent risk score, prospective risk score, unique acute diagnoses, chronic condition diagnoses, inpatient claimants, or member count.

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

inputting, to the first machine learning model, target data associated with the first target and feature data associated with the one or more features;

generating, by the first machine learning model, an adjusted target value based on the target data and the feature data;

calculating an error by comparing the adjusted target value to an actual target value; and

updating one or more weights of the first machine learning model using the calculated error.

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

receiving, at a server and from a data provider system, a record comprising the first target and the second target, wherein the record is in a non-standardized format;

extracting the first and second targets from the record;

converting the first and second targets into a numerical vector; and

storing the converted first and second targets as a numerical matrix.

9 . A system, comprising:

a memory; and

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

select a first target, wherein the first target includes a first target value;

in real-time and based on selection of the first target:

retrieve a first dataset from a plurality of data sources, the first dataset comprising one or more features associated with the first target;

generate a first external weight by inputting the first target value and the one or more features associated with the first target into a first machine learning model corresponding to the first target, wherein the first machine learning model is configured to output an impact of the one or more features on the first target value;

generate a first adjusted target value by combining the first target value with the first external weight;

determine a first error value based on the first target value and the first adjusted target value; and

normalize the first error value via a statistical measure;

select a second target, wherein the second target includes a second target value;

in real-time and based on selection of the second target:

retrieve a second dataset from the plurality of data sources, the second dataset comprising one or more features associated with the second target;

generate a second external weight by inputting the second target value and the one or more features associated with the second target into a second machine learning model corresponding to the second target, wherein the second machine learning model is configured to output an impact of the one or more features on the second target value;

generate a second adjusted target value by combining the second target value with the second external weight;

determine a second error value based on the second target value and the second adjusted target value; and

normalize the second error value via the statistical measure; and

in response to normalization of the second error value or normalization of the first error value:

generate an index value based on the first normalized error value and the second normalized error value; and

transmit, based on the index value being within a threshold percentile of a plurality of index values, an alert to one or more user computing devices.

10 . The system of claim 9 , wherein the index value is generated based on the average of the first normalized error value and the second normalized error value.

11 . The system of claim 9 , wherein the at least one processor is further configured to:

in response to generating the index value, automatically generate a message including the index value, the first target, and the second target; and

transmit the message via a network to the one or more user computing devices.

12 . The system of claim 11 , wherein the message comprises a link to access a graphical user interface (GUI) via the network, wherein the link is unique to the generated index value.

13 . The system of claim 9 , wherein the statistical measure is a z-score.

14 . The system of claim 9 , wherein the first and second machine learning models include one or more weights corresponding to average age, percentage male, percentage female, continuously enrolled months, concurrent risk score, prospective risk score, unique acute diagnoses, chronic condition diagnoses, inpatient claimants, or member count.

15 . The system of claim 9 , wherein the at least one processor is further configured to:

input, to the first machine learning model, target data associated with the first target and feature data associated with the one or more features;

generate, by the first machine learning model, an adjusted target value based on the target data and the feature data;

calculate an error by comparing the adjusted target value to an actual target value; and

update one or more weights of the first machine learning model using the calculated error.

16 . 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:

selecting a first target, wherein the first target includes a first target value;

in real-time and based on selecting the first target:

retrieving a first dataset from a plurality of data sources, the first dataset comprising one or more features associated with the first target;

generating a first external weight by inputting the first target value and the one or more features associated with the first target into a first machine learning model corresponding to the first target, wherein the first machine learning model is configured to output an impact of the one or more features on the first target value;

generating a first adjusted target value by combining the first target value with the first external weight;

determining a first error value based on the first target value and the first adjusted target value; and

normalizing the first error value via a statistical measure;

selecting a second target, wherein the second target includes a second target value;

in real-time and based on selecting the second target:

retrieving a second dataset from the plurality of data sources, the second dataset comprising one or more features associated with the second target;

generating a second external weight by inputting the second target value and the one or more features associated with the second target into a second machine learning model corresponding to the second target, wherein the second machine learning model is configured to output an impact of the one or more features on the second target value;

generating a second adjusted target value by combining the second target value with the second external weight;

determining a second error based on the second target value and the second adjusted target value; and

normalizing the second error value via the statistical measure; and

in response to normalizing the second error value or normalizing the first error value:

generating an index value based on the first normalized error value and the second normalized error value; and

transmitting, based on the index value being within a threshold percentile of a plurality of index values, an alert to one or more user computing devices.

17 . The non-transitory computer-readable device of claim 16 , wherein the first and second machine learning models include one or more weights corresponding to average age, percentage male, percentage female, continuously enrolled months, concurrent risk score, prospective risk score, unique acute diagnoses, chronic condition diagnoses, inpatient claimants, or member count.

18 . The non-transitory computer-readable device of claim 16 , wherein the index value is generated based on the average of the first normalized error value and the second normalized error value.

19 . The non-transitory computer-readable device of claim 16 , the operations further comprising:

in response to generating the index value, automatically generating a message including the index value, the first target, and the second target; and

transmitting the message via a network to the one or more user computing devices,

wherein the message comprises a link to access a graphical user interface (GUI) via the network, and

wherein the link is unique to the generated index value.

20 . The non-transitory computer-readable device of claim 16 , wherein the statistical measure is a z-score.