CONTEXT

Established nearly 70 years ago, Dr Lal PathLabs has come a long way and is now one of the top pathology labs in India. With over 3,368 diagnostic tests and services offered, they are capable of performing virtually all of the diagnostic healthcare tests and services currently prescribed by physicians in India. They have earned the reputation of being amongst the most trustworthy and reliable pathology labs of India, having delivered the most accurate reports over the years. Their accuracy, however, didn’t remain confined to their pathology labs. They wanted to cater to their customers better by giving them an accurate time by when they could expect their medical reports, saving them the hassle of frequent inquiry and elevating the response process from manual to automated.

We, at [x]cube LABS, knew that the solution lay in predicting outcomes to the best accuracy. Our expert team, having extensive experience on a range of machine learning tasks, accessed their historic and master data, cleaned, and analyzed it. What came next was a customized algorithm that was a perfect fit for the client’s needs.

Dr Lal Pathlabs identified that a pain point for their customers was having to wait for their reports to be delivered. They would often make calls inquiring about it or visit the lab without inquiring, only to be given a tentative time of report delivery which was subject to various unforeseen circumstances. The diagnosis sometimes involved multiple tests across different departments such as hematology, neurology, etc which demanded strong coordination and communication among them. This made predicting an exact delivery time even more difficult.

The prediction was dependent on multiple factors that were closely interconnected and influenced each other in more than one way. We had to analyze a complex structure to understand the connection between multiple tests that fall under the same group as well as different groups under different technical departments. It was a complex algorithm but taking multiple parameters into account in order to give the consumers an accurate result entailed it be so. Our team used a comprehensive approach where algorithms were not only customized but applied multiple times to gain the accuracy that we had hoped for.

We provided a solution that was able to closely monitor and derive the time of report delivery. It took into account the time consumed in various phases of the lab and had scope to include exceptional delays due to reasons such as machine breakdown. The estimated time came with a margin of two hours, which could be a minus or plus due to unexpected human interventions.

KEY FEATURES

• High accuracy: The algorithms combined the power of Random Forest and Gradient Booster algorithms for higher accuracy.
• Individual Test estimate: While a consolidated report is delivered to the customer, it is possible to see the individual time for each test as well.
• Induced Delays: The solution has scope to accommodate exceptional delays that are caused by reasons such as instrument breakdown, manpower absence, logistics delay.
• Configurable Options: It is possible to add buffer time, exception list, and stepwise release to selected labs.

The solution has helped hundreds of customers track their report status online. They can check it on the website, IVR, collection centers inquiry, or through the mobile application. Received well by the client and their customers alike, the solution has eliminated the inconvenience caused to both the parties by the uncertainty of manual, speculative predictions.