PDPM Success Story at SNF
Helping Skilled Nursing Facilities (SNF) can make the transition to Patient Driven Payment Model (PDPM) in a painless and patient first manner.
An Overview of the Change from RUG IV to PDPM
SNF’s in the United States is in the midst of one of the most significant transformations in recent times. An SNF is an in-patient facility where patients stay after their hospital visits. Until recently, the SNF’s billed and were reimbursed based on the total therapy minutes they expended on a patient. Under this model, the Federal government was concerned about the incentive to stretch the therapy process out at the SNF. As the Federal government reimburses the SNFs for some of the costs, they wanted to see a more cost-effective approach being used by the SNFs. The current SNF Prospective Payment System (PPS) was based on Resource Utilization Groups (RUGS).
RUG-IV is a patient classification system for SNFs used by the Federal government to determine reimbursement levels. Payments are determined by grouping the patients into groups based on their resource needs, of which the therapy minutes make up a large chunk. To address this, the Federal government rolled out PDPM that would focus on classifying patients into patient groups based on specific data-driven patient characteristics. The initial deadline for the utilization of PDPM was October 1, 2019, but many SNF has failed to meet it.
While the intention of the Federal government was noble, the reality at the SNF was not fully taken into account. Only 11 percent of acute care providers use an integrated electronic health record (EHR), and only two percent of acute care and LTPAC providers are using IT-driven strategies only to coordinate patient care and transfer data.
Reference link to all regulations required to be in-compliance for PDPM : https://www.cms.gov/Medicare/Medicare-Fee-for-Service-Payment/SNFPPS/List-of-SNF-Federal-Regulations
PDPM Custom Algorithm Case Study
MindTrades consulting has been in the Big Data, Machine Learning, and Artificial Intelligence space for some time. Its experience in big business customer marketing was predicated on finding similar customers and classifying them into clusters or similar groups. In 2019 it decided to apply its learnings to a multi-location SNF locations in the United States.
A total of 21 patient parameters that were thought to be relevant were identified as the input to the algorithm. All patient data going back three years from multiple sources and formats was ingested into a single platform. This included medical imaging too. The algorithm was then trained and tested with a standard 80: 20 rule, where 80% of the data was used for training the algorithm, and it was then tested on 20% of the data.
Results of the Algorithm for PDPM
The custom predictive algorithm written for this client resulted in a 91.6% accuracy level. Predictions were returned in under a second to the physician who, of course still had the final say in determining the patient care decisions. The entire solution was HIPPA compliant.
An Overview of the Technical Solution
The Mindtrades solution can ingest data from multiple sources, including non-digital formats. The database then created can be stored at either client premises or in a certified and secure cloud solution.
Learn how a multi-location Skilled Nursing Facility had a custom predictive algorithm built achieving an accuracy rate of 91.6% within 90 days