#420: Cannabis and Machine Learning, a Joint Venture
Cannabis growers and sellers are rolling in and cashing out on machine learning Regardless of scale, cannabis growers and sellers are...
The client is one of the largest home-based care organizations nationwide. The company launched in 2018 as a merger between three respected home healthcare providers. At the time of the merger, it already served more than 65,000 patients daily and employed about 32,500 caregivers.
The three-way merger had a clear focus of building a longitudinal care platform, which could serve patients across the whole care continuum. At the core of this platform were modern tech products that were meant to efficiently ingest, process, and leverage healthcare data.
At the same time, the three-way merger meant that IT workloads and operations needed to move to the cloud. The variety of internal services powering each of the 3 companies made the migration towards the cloud particularly complex: our client found itself having to build an enterprise cloud environment from scratch.
The client performs referral processing across a large referral network, which is based on a variety of data sources. Different parts of the business use different APIs and software tools for gathering and analyzing referral data.
The first challenge that needed to be tackled was to identify and align the different internal processing workflows. The client needed to gain a better understanding of their infrastructure needs, in order to make referral data accessible in a unified, standardized format. Only a well-defined infrastructure backbone would allow further automation of referral processing.
Secondly, the client required a way to speed up referral processing and to make sure that it is done accurately. The limitations of their legacy system prevented them from making fast and accurate referral assessments. The lack of a standardized infrastructure led to large referral processing times, up to a few hours. However, the client’s long-term goal is to minimize the decision-making process, bringing it down to a few minutes, in order to stay relevant in the competitive referral market.
Goal: Increase speed of referral process, from 2 hours to 2 minutes.
Lastly, the client needed to cut down on time-consuming, error-prone, and inefficient processing. They needed a way to minimize the manual labor involved in referral assessments and to reduce the number of software tools needed in the process. Not only would this make them more competitive, but it would enable them to cut down on operational costs.
Blue Orange Digital implemented a cloud-based solution that would allow the client to more easily handle referral data. This is a long-term task, involving more than simply setting up a data pipeline and integrating different data sources. The goal is to understand and redesign the client’s internal processes and workflows, in order to support their most important goal: automated referral processing.
The workload was broken down into different stages:
1. The team identified different data sources and the best ways to handle them
Handling different data sources must not be limited to the performance of manual human processing. Instead, automated processes can be established, which speed up decision times and minimize errors. With data involved in referral processing being quite diverse, the team has identified different types and data sources. Some examples include:
Data from referral attachments, containing case information, diagnosis, medication
Metadata of regional branches, informing about capacity, scheduling, staff availability
All this data comes in different formats and over a variety of channels. In the long term, this data should be available as API data, in a unified format. In order to get there, the team is implementing proof of concept modules that allows them to test the potential for automation.
For example, part of the solution is a web scraping module, which automatically checks referral portals for new referrals and makes it possible to download that data in a predefined format (for example as a PDF). That data can be further processed and integrated with the other data sources as part of the referral decision process.
2. The team identified a cloud environment that best fits the client’s needs
The proposed solution is developed in the Azure environment, where proven tools can be used across the whole data processing pipeline: from data collection, ingestion, and extraction to modeling and analytics. Azure API Management makes it possible to expose and connect to data from both internal and external services. At the same time, Azure Functions enable the team to define rules and triggers for capturing relevant data events. Last but not least, Azure Form Recognizer enables them to extract structured data from form documents. The established ecosystem gives the team great flexibility in mapping internal business processes to data processing tools.
Another benefit of the established cloud environment is that of data security and privacy. Since Microsoft Azure offers a wide range of services which are HITRUST CSF Certified, the infrastructure is designed to respect all regulatory requirements and industry best practices.
Blue Orange Digital enabled the client to plan and implement a cost-effective infrastructure that delivers the necessary performance for their use case. By keeping track of their long-term goals, the Solutions Architects can leverage proven cloud tools to account for their most urgent needs.
The new infrastructure enables them to stand out in the competitive market of referral bidding. This is achieved by making the best of automation, cutting down on manual labor costs, and minimizing inefficient processes. The home healthcare referral process is being reinvented in the cloud and gets a well-deserved digital makeover.