ST IT CLOUD Case Study: Assertive Machine Learning
In this article, we will discuss the case study from the healthcare sector, whose main challenge was processing a large volume of clients, causing bottlenecks of high complexity regarding medical auditing in exam and procedure authorizations. ST IT Cloud's work was to develop, through Data Lake and Machine Learning, a system that operated in an Orchestrated and Economical manner.
THE ST IT CHALLENGE
As previously mentioned, information processing within Unimed-BH was becoming extremely complex when it came to approving medical exams and authorizing forms. Because of this, IT superintendent Ezequiel Ribeiro da Silva explains that currently more than 180,000 procedure requests are generated per month. "About 145,000 of these requests are already handled by Unimed-BH's systems through fixed rules. However, there are 35,000 monthly requests that are forwarded to medical auditors, whose function is to evaluate these requests case by case", he says.
With this, three main points were identified as necessary to automate the process:
- Reduce expenses with medical auditing in exam and procedure requests
- Make evaluations for approval and auditing more effective
- Streamline the exam approval process
DISCOVER THE ST IT SOLUTIONS
Development of Architecture and a system capable of loading structured and semi-structured, transactional and historical data and files into a Data Lake in an Orchestrated and Economical manner.
Additionally, a Machine Learning model was developed that was assertive enough to automatically authorize medical exams and procedures, with validation and display of information and alerts in Near Real Time.
"The current solution is the result of various experiments and different architecture designs in different pilots. The flexibility of using different services available on AWS, with the per-allocation billing system, allowed Unimed-BH to experiment with various scenarios before reaching the most suitable one", reveals Silva to AWS.
Currently, the project is showing good results, with 40% faster processes, and we intend to improve this to 60% as the model matures, further enhancing the experience and results!
If your company has a data, innovation, or cloud migration challenge, contact our specialists.
