System and method for record linkage
Methods, systems, and computer-readable media are provided for facilitating record matching and entity resolution and for enabling improvements in record linkage. A power-spectrum-based temporal pattern-specific weight may be incorporated into record linkage methods to enhance the record linkage accuracy and statistical performance. For example, in embodiments, a value-specific weight may be calculated from a population-based frequency of field-specific values and provides an opportunity to capture and measure the relative importance of specific values found in a field. A timeseries-derived Bayesian power spectrum weight may be calculated from the population-based frequency of temporal pattern-specific values in terms of intensities at various frequencies of the power spectrum computed from the timeseries, and further provides an opportunity to capture and measure the relative importance of specific sequences of care episodes.
1. A computer-implemented method comprising:
receiving a target record from a first record system, the target record comprising one or more first date-time variables associated with a target patient, wherein each of the one or more first date-time variables identifies a date and/or time;
receiving a candidate record from a second record system, the candidate record comprising a second date-time variable for an episode associated with the target patient, wherein the second date-time variable identifies a date and/or time;
generating an interim record that aggregates the first date-time variables with the second date-time-variable;
generating a power spectrum based on the interim record;
predicting, based on at least one demographic characteristic of a subject, a record linkage weight based on a similarity of a demographic variable associated with the candidate record and the target record;
determining, based on the record linkage weight and the power spectrum, that the candidate record is related to the target record; and
storing an indication that the candidate record is to be linked to the target record.
2. The computer-implemented method of claim 1 , further comprising:
determining a power spectra weight based on a characteristic of the power spectrum and a corresponding reference characteristic generated using a set of reference power spectra, wherein determining that the candidate record is related to the target record is based on the record linkage weight and the power spectra weight.
3. The computer-implemented method of claim 2 , further comprising:
determining a composite weight of the record linkage weight and the power spectra weight, wherein determining that the candidate record is related to the target record comprises determining that the composite weight satisfies a threshold.
4. The computer-implemented method of claim 2 , wherein determining the power spectra weight comprises:
determining a likelihood value of each spectrum of the set of reference power spectra; and
normalizing the determined likelihood values.
5. The computer-implemented method of claim 2 , wherein the set of reference power spectra are generated using a Bayesian Markov Chain Monte Carlo simulation.
6. The computer-implemented method of claim 1 , wherein the second date-time variable comprises a date-time value associated with a most recent episode stored in association with the candidate record.
7. The computer-implemented method of claim 1 , wherein the first and second record systems comprise the same record system.
8. The computer-implemented method of claim 1 , further comprising:
presenting, via an interface, an indication that the candidate record is related to the target record.
9. A system comprising:
one or more data processors; and
a non-transitory computer readable storage medium containing instructions which, when executed on the one or more data processors, cause the one or more data processors to perform a set of actions including:
receiving a target record from a first record system, the target record comprising one or more first date-time variables associated with a target patient, wherein each of the one or more first date-time variables identifies a date and/or time;
receiving a candidate record from a second record system, the candidate record comprising a second date-time variable for an episode associated with the target patient, wherein the second date-time variable identifies a date and/or time;
generating an interim record that aggregates the first date-time variables with the second date-time-variable;
generating a power spectrum based on the interim record;
predicting, based on at least one demographic characteristic of a subject, a record linkage weight based on a similarity of a demographic variable associated with the candidate record and the target record;
determining, based on the record linkage weight and the power spectrum, that the candidate record is related to the target record; and
storing an indication that the candidate record is to be linked to the target record.
10. The system of claim 9 , wherein the set of actions further includes:
determining a power spectra weight based on a characteristic of the power spectrum and a corresponding reference characteristic generated using a set of reference power spectra, wherein determining that the candidate record is related to the target record is based on the record linkage weight and the power spectra weight.
11. The system of claim 10 , wherein the set of actions further includes:
determining a composite weight of the record linkage weight and the power spectra weight, wherein determining that the candidate record is related to the target record comprises determining that the composite weight satisfies a threshold.
12. The system of claim 10 , wherein determining the power spectra weight comprises:
determining a likelihood value of each spectrum of the set of reference power spectra; and
normalizing the determined likelihood values.
13. The system of claim 10 , wherein the set of reference power spectra are generated by sampling permutations of the power spectrum.
14. The system of claim 9 , wherein the second date-time variable comprises a date-time value associated with a most recent episode stored in association with the candidate record.
15. The system of claim 9 , wherein the first and second record systems comprise the same record system.
16. The system of claim 9 , wherein the set of actions further includes:
presenting, via an interface, an indication that the candidate record is related to the target record.
17. A computer-program product tangibly embodied in a non-transitory machine-readable storage medium, including instructions configured to cause one or more data processors to perform a set of actions including:
receiving a target record from a first record system, the target record comprising one or more first date-time variables associated with a target patient, wherein each of the one or more first date-time variables identifies a date and/or time;
receiving a candidate record from a second record system, the candidate record comprising a second date-time variable for an episode associated with the target patient, wherein the second date-time variable identifies a date and/or time;
generating an interim record that aggregates the first date-time variables with the second date-time-variable;
generating a power spectrum based on the interim record;
predicting, based on at least one demographic characteristic of a subject, a record linkage weight based on a similarity of a demographic variable associated with the candidate record and the target record;
determining, based on the record linkage weight and the power spectrum, that the candidate record is related to the target record; and
storing an indication that the candidate record is to be linked to the target record.
18. The computer-program product of claim 17 , wherein the set of actions further includes:
determining a power spectra weight based on a characteristic of the power spectrum and a corresponding reference characteristic generated using a set of reference power spectra, wherein determining that the candidate record is related to the target record is based on the record linkage weight and the power spectra weight.
19. The computer-program product of claim 18 , wherein the set of actions further includes:
determining a composite weight of the record linkage weight and the power spectra weight, wherein determining that the candidate record is related to the target record comprises determining that the composite weight satisfies a threshold.
20. The computer-program product of claim 18 , wherein determining the power spectra weight comprises:
determining a likelihood value of each spectrum of the set of reference power spectra; and
normalizing the determined likelihood values.