Engagement overview

A leading South African hospital group, which processes medical claims with fund administrators via a business to business (B2B) process, found a high volume of claims were being rejected by administrators for various reasons relating to patient admission, billing or other clinical factors. This ultimately resulted in short payment of claims, costing the group millions each month.
The high volume of claims processed made it difficult to identify patterns in rejections over time, meaning “trends” were identified based on intuition and manual interpretation. The manual nature of this process resulted in long from trend origination to identification lead times, resulting in delayed reactions – or no action at all – and revenue lost as a result of claims rejections not being resolved.

BSG proposed a unique multi-layered, data-led approach to implement a solution to identify, and proactively monitor trends across the business:

Statistical methods, adapted from engineering processes, were applied to develop the trends and flag significant shifts, in near-real-time. The cause of the shift was also identified using hypothesis testing, as well as the expected impact if no action was taken to adjust the approach. Both negative and positive movements were monitored, and the key results communicated to stakeholders via email and an interactive dashboard.
Solution architecture was done using Microsoft Azure cloud technology, leveraging Azure Data Factory and Azure Machine Learning (using Python) to process, transform and model data. This pipeline processes the data, outputting significant results via an email summary, and a detailed view in Microsoft PowerBI. The PowerBI dashboard could be used to “slice and dice” data in order to understand rejection patterns from a variety of different perspectives.
The deployed solution monitors over 650,000 unique trends on a weekly basis. Trends relate to specific funder rejections, across a number of characteristics unique to each case billed and submitted for payment. These characteristics could include specifics of the medical aid plan in relation to that specific hospital, the attending doctor, and clinical aspects of the case that could influence a trend in payment rejections.

Making a difference

BSG successfully delivered a solution that will continuously monitor for significant shifts in rejection trends and provide a view of the potential impact of inaction. In the time since implementation*, over R18m in short-payment rejections have been identified and actioned, resulting in a significant reduction in working capital costs, and in the cost of manual intervention.
* 7 months at time of publishing