Focusing unstructured data and generating focused data determinations from an unstructured data set
Each year, almost 10% of claims are denied by payers (i.e., health insurance plans). With the cost to recover these denials and underpayments, predicting payer response (likelihood of payment) from claims data with a high degree of accuracy and precision is anticipated to improve healthcare staffs' performance productivity and drive better patient financial experience and satisfaction in the revenue cycle (Barkholz, 2017). However, constructing advanced predictive analytics models has been considered challenging in the last twenty years. That said, we propose a (low-level) context-dependent compact representation of patients' historical claim records by effectively learning complicated dependencies in the (high-level) claim inputs. Built on this new latent representation, we demonstrate that a deep learning-based framework, Deep Claim, can accurately predict various responses from multiple payers using 2,905,026 de-identified claims data from two US health systems. Deep Claim's improvements over carefully chosen baselines in predicting claim denials are most pronounced as 22.21% relative recall gain (at 95% precision) on Health System A, which implies Deep Claim can find 22.21% more denials than the best baseline system.
1. A computer-implemented method comprising:
integrating, by one or more processors, a first model into a second model by connecting one or more layers of the first model to one or more layers of the second model, wherein the first model has been trained based at least in part on a first data set and the first data set is associated with a first model domain;
training, by the one or more processors, a first portion of the second model based at least in part on the first data set;
freezing, by the one or more processors, the first model integrated into the second model; and
training, by the one or more processors, a second portion of the second model based at least in part on a second data set, wherein the second data set is associated with a second model domain that is different than the first model domain.
2. The computer-implemented method of claim 1 , further comprising:
unfreezing the first model integrated into the second model; and
fine-tuning the second model based at least in part on the second data set.
3. The computer-implemented method of claim 1 , further comprising:
storing the second model.
4. The computer-implemented method of claim 1 , further comprising:
receiving input data associated with an unstructured data set; and
generating, using the second model, output data based at least in part on the input data.
5. The computer-implemented method of claim 1 , further comprising:
increasing a learning rate of the second model upon freezing the first model.
6. The computer-implemented method of claim 1 , further comprising:
unfreezing the first model integrated into the second model; and
decreasing the learning rate of the second model.
7. The computer-implemented method of claim 1 , wherein the first data set comprises waste and error data and wherein the second data set comprises fraud data.
8. A system comprising memory and one or more processors communicatively coupled to the memory, the one or more processors configured to: integrate a first model into a second model by connecting one or more layers of the first model to one or more layers of the second model, wherein the first model is initially trained based at least in part on a first data set that is associated with a first model domain;
train a first portion of the second model based at least in part on the first data set;
freeze the first model integrated into the second model; and
train a second portion of the second model based at least in part on a second data set that is associated with a second model domain different than the first model domain.
9. The computing system of claim 8 , wherein the instructions further configure the apparatus to:
unfreeze the first model integrated into the second model; and
fine-tune the second model based at least in part on the second data set.
10. The computing system of claim 8 , wherein the instructions further configure the apparatus to:
store the second model.
11. The computing system of claim 8 , wherein the instructions further configure the apparatus to:
receive input data associated with an unstructured data set; and
generate, using the second model, output data based at least in part on the input data.
12. The computing system of claim 8 , wherein the instructions further configure the apparatus to:
increase a learning rate of the second model upon freezing the first model.
13. The computing system of claim 8 , wherein the instructions further configure the apparatus to:
unfreeze the first model integrated into the second model; and
decrease the learning rate of the second model.
14. The computing system of claim 8 , wherein the first data set comprises a waste & error data and wherein the second data set comprises fraud data.
15. One or more non-transitory computer-readable storage media including instructions that, when executed by one or more processors, cause the one or more processors to:
integrate a first model into a second model by connecting one or more layers of the first model to one or more layers of the second model, wherein the first model is initially trained based at least in part on a first data set that is associated with a first model domain;
train a first portion of the second model based at least in part on the first data set;
freeze the first model integrated into the second model; and
train a second portion of the second model based at least in part on a second data set that is associated with a second model domain different than the first model domain.
16. The one or more non-transitory computer-readable storage media of claim 15 , wherein the one or more processors are further caused to:
unfreeze the first model integrated into the second model; and
fine-tune the second model based at least in part on the second data set.
17. The one or more non-transitory computer-readable storage media of claim 15 , wherein the one or more processors are further caused to:
store the second model.
18. The one or more non-transitory computer-readable storage media of claim 15 , wherein the one or more processors are further caused to:
receive input data associated with an unstructured data set; and
generate, using the second model, output data based at least in part on the input data.
19. The one or more non-transitory computer-readable storage media of claim 15 , wherein the one or more processors are further caused to:
increase a learning rate of the second model upon freezing the first model.
20. The one or more non-transitory computer-readable storage media of claim 19 , wherein the one or more processors are further caused to:
unfreeze the first model integrated into the second model; and
decrease the learning rate of the second model.